Compare Mistral Large 3 and Mistral Medium 3: European Sovereign AI Flagship vs Mid-Tier Workhorse
Mistral Large 3 vs Mistral Medium 3: which should I use?
Start with Mistral Medium when its exact SKU, host, quota, concurrency, and accepted-result floor are proven. Escalate to Large only on a measured capacity or validity trigger, with bounded retries and a safe fallback. Verified 2026-09-02; unmatched hosts, latency, and acceptance denominators remain Unavailable.
Verified 2026-09-02
Where can you find price, speed, and task evidence for Mistral Large 3 and Mistral Medium 3?
Which tasks fit Mistral Large 3 and Mistral Medium 3?
Best-fit task signals from the published model strengths; this is not a substitute for a controlled benchmark.
| Fact | Mistral Large 3 | Mistral Medium 3 |
|---|---|---|
| Best fit | Strong multimodal reasoning and coding at a lower cost than the big-lab flagships. | Balanced production workloads that need Mistral’s current frontier tier. |
| Reasoning mode | Unavailable | Unavailable |
What does a Coding Agent workload cost with Mistral Large 3 and Mistral Medium 3?
Modeled for Coding Agent: 20,000 input + 2,000 output tokens per task, 1 turn. This is a workload model, not a provider quote.
| Fact | Mistral Large 3 | Mistral Medium 3 |
|---|---|---|
| Coding Agent / task | $0.01 (modeled) (winner) | $0.04 (modeled) |
| Input / output rate | $0.50 / $1.50 per M | $1.50 / $7.50 per M |
How fast are Mistral Large 3 and Mistral Medium 3?
Only non-estimated benchmark results are shown as measured.
| Fact | Mistral Large 3 | Mistral Medium 3 |
|---|---|---|
| Measured throughput | 61 tokens/s | 92 tokens/s (winner) |
| Time to first token | 400 ms | 320 ms |
How compatible are Mistral Large 3 and Mistral Medium 3 with APIs?
Provider-level wire-format and SDK facts; model-specific parameter differences may still apply.
| Fact | Mistral Large 3 | Mistral Medium 3 |
|---|---|---|
| OpenAI SDK | Usable | Usable |
| Request shape | OpenAI-compatible chat/completions endpoint at api.mistral.ai/v1. | OpenAI-compatible chat/completions endpoint at api.mistral.ai/v1. |
| Streaming | SSE (OpenAI delta) | SSE (OpenAI delta) |
What are the privacy and retention policies for Mistral Large 3 and Mistral Medium 3?
Provider policy and residency facts from the verified provider registry. A missing retention policy is not treated as a guarantee.
| Fact | Mistral Large 3 | Mistral Medium 3 |
|---|---|---|
| Provider says API data trains models | No | No |
| Published retention period | Unavailable | Unavailable |
| Data residency | EU-hosted by default | EU-hosted by default |
How much effort does it take to migrate between Mistral Large 3 and Mistral Medium 3?
Direction is from each displayed model to the other. Effort is derived from provider compatibility and published parameter maps.
| Fact | Mistral Large 3 | Mistral Medium 3 |
|---|---|---|
| Move Mistral Large 3 → Mistral Medium 3 | drop-in; 0 breaking parameter differences | Target: Mistral Medium 3 |
| Move Mistral Medium 3 → Mistral Large 3 | Source: Mistral Medium 3 | drop-in; 0 breaking parameter differences |
| Why | Same provider; only the model identifier changes. | Same provider; only the model identifier changes. |
Batch 55 · exact-pair operating architecture contributions. Surface verification: 2026-09-02. Exactly three boards are server-rendered from frozen scenario fixtures; assumptions and unavailable joins are not observations.
Exact pair boundary: Mistral; immutable SKU, host, account, concurrency, quota, validator, and accepted-result denominator must join. Requested models are mistral-large and mistral-medium. Provider, rate, pricing, and task pages remain fact owners.
Immutable Large/Medium SKU resolver
Frozen Batch 55 scenario board. Formula / deterministic rule: eligible = immutable SKU + exact host/account + effective ID + lifecycle evidence; alias or host drift => block Boundary: Owns same-provider Large/Medium identity and capacity admission, not Mistral platform policy or pricing alone.
| Frozen scenario / field ID | Pair, identity, host, surface, and evidence fields | Result | State |
|---|---|---|---|
batch55-mistral-large-vs-mistral-medium-m1-r1direct Large SKU | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai or declared host; hostB=api.mistral.ai or declared host; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=direct Large SKU; requested/effective ID; immutable SKU; host/account; region; lifecycle date; schema/tool/modality contract; current/stale/ambiguous state; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-02; measurement versus assumption=explicit | Unavailable — direct Large SKU is a deterministic fixture pending exact-ID, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch55-mistral-large-vs-mistral-medium-m1-r2direct Medium SKU | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai or declared host; hostB=api.mistral.ai or declared host; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=direct Medium SKU; requested/effective ID; immutable SKU; host/account; region; lifecycle date; schema/tool/modality contract; current/stale/ambiguous state; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-02; measurement versus assumption=explicit | Unavailable — direct Medium SKU is a deterministic fixture pending exact-ID, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch55-mistral-large-vs-mistral-medium-m1-r3gateway alias | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai or declared host; hostB=api.mistral.ai or declared host; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=gateway alias; requested/effective ID; immutable SKU; host/account; region; lifecycle date; schema/tool/modality contract; current/stale/ambiguous state; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-02; measurement versus assumption=explicit | Unavailable — gateway alias is a deterministic fixture pending exact-ID, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch55-mistral-large-vs-mistral-medium-m1-r4regional host | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai or declared host; hostB=api.mistral.ai or declared host; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=regional host; requested/effective ID; immutable SKU; host/account; region; lifecycle date; schema/tool/modality contract; current/stale/ambiguous state; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-02; measurement versus assumption=explicit | Unavailable — regional host is a deterministic fixture pending exact-ID, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch55-mistral-large-vs-mistral-medium-m1-r5deprecated SKU | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai or declared host; hostB=api.mistral.ai or declared host; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=deprecated SKU; requested/effective ID; immutable SKU; host/account; region; lifecycle date; schema/tool/modality contract; current/stale/ambiguous state; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-02; measurement versus assumption=explicit | Unavailable — deprecated SKU is a deterministic fixture pending exact-ID, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch55-mistral-large-vs-mistral-medium-m1-r6unknown effective SKU | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai or declared host; hostB=api.mistral.ai or declared host; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=unknown effective SKU; requested/effective ID; immutable SKU; host/account; region; lifecycle date; schema/tool/modality contract; current/stale/ambiguous state; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-02; measurement versus assumption=explicit | Unavailable — unknown effective SKU has no matched, dated bilateral observation. | FAIL CLOSED — manual or probe evidence required |
First-party provenance: Mistral model documentation; verification date 2026-09-02. Missing or conflicting joins fail closed.
Concurrency-to-validity frontier receipt
Frozen Batch 55 scenario board. Formula / deterministic rule: frontier = accepted results per minute at declared concurrency with matched host/config/validator; unmatched host => Untested Boundary: Owns exact-pair capacity and accepted-result frontier; broad speed rankings remain elsewhere.
| Frozen scenario / field ID | Pair, identity, host, surface, and evidence fields | Result | State |
|---|---|---|---|
batch55-mistral-large-vs-mistral-medium-m2-r1single request | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai or declared host; hostB=api.mistral.ai or declared host; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=single request; workload/config hashes; exact SKU/host; concurrency; TTFT/throughput/error observations or Untested; validator/denominator; accepted results per minute; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-02; measurement versus assumption=explicit | Unavailable — single request is a deterministic fixture pending exact-ID, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch55-mistral-large-vs-mistral-medium-m2-r2interactive 10-way | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai or declared host; hostB=api.mistral.ai or declared host; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=interactive 10-way; workload/config hashes; exact SKU/host; concurrency; TTFT/throughput/error observations or Untested; validator/denominator; accepted results per minute; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-02; measurement versus assumption=explicit | Unavailable — interactive 10-way is a deterministic fixture pending exact-ID, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch55-mistral-large-vs-mistral-medium-m2-r3sustained 50-way | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai or declared host; hostB=api.mistral.ai or declared host; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=sustained 50-way; workload/config hashes; exact SKU/host; concurrency; TTFT/throughput/error observations or Untested; validator/denominator; accepted results per minute; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-02; measurement versus assumption=explicit | Unavailable — sustained 50-way is a deterministic fixture pending exact-ID, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch55-mistral-large-vs-mistral-medium-m2-r4burst 200-way | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai or declared host; hostB=api.mistral.ai or declared host; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=burst 200-way; workload/config hashes; exact SKU/host; concurrency; TTFT/throughput/error observations or Untested; validator/denominator; accepted results per minute; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-02; measurement versus assumption=explicit | Unavailable — burst 200-way is a deterministic fixture pending exact-ID, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch55-mistral-large-vs-mistral-medium-m2-r5batch queue | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai or declared host; hostB=api.mistral.ai or declared host; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=batch queue; workload/config hashes; exact SKU/host; concurrency; TTFT/throughput/error observations or Untested; validator/denominator; accepted results per minute; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-02; measurement versus assumption=explicit | Unavailable — batch queue is a deterministic fixture pending exact-ID, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch55-mistral-large-vs-mistral-medium-m2-r6throttled response | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai or declared host; hostB=api.mistral.ai or declared host; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=throttled response; workload/config hashes; exact SKU/host; concurrency; TTFT/throughput/error observations or Untested; validator/denominator; accepted results per minute; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-02; measurement versus assumption=explicit | Unavailable — throttled response is a deterministic fixture pending exact-ID, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
First-party provenance: Mistral API rate limits; verification date 2026-09-02. Missing or conflicting joins fail closed.
Medium-first capacity-overflow board
Frozen Batch 55 scenario board. Formula / deterministic rule: escalate = evidenced Medium trigger + admitted Large + bounded retry + settled side effects; repeated failure => manual Boundary: Owns Medium-first routing and overflow for this pair, preserving Medium where it remains safe.
| Frozen scenario / field ID | Pair, identity, host, surface, and evidence fields | Result | State |
|---|---|---|---|
batch55-mistral-large-vs-mistral-medium-m3-r1Medium-only | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai or declared host; hostB=api.mistral.ai or declared host; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=Medium-only; trigger evidence; queue/retry state; tokens carried forward; side-effect class; Large eligibility; attempt cap; route/hold/manual outcome; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-02; measurement versus assumption=explicit | Unavailable — Medium-only is a deterministic fixture pending exact-ID, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch55-mistral-large-vs-mistral-medium-m3-r2context overflow | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai or declared host; hostB=api.mistral.ai or declared host; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=context overflow; trigger evidence; queue/retry state; tokens carried forward; side-effect class; Large eligibility; attempt cap; route/hold/manual outcome; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-02; measurement versus assumption=explicit | Unavailable — context overflow is a deterministic fixture pending exact-ID, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch55-mistral-large-vs-mistral-medium-m3-r3latency-SLA breach | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai or declared host; hostB=api.mistral.ai or declared host; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=latency-SLA breach; trigger evidence; queue/retry state; tokens carried forward; side-effect class; Large eligibility; attempt cap; route/hold/manual outcome; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-02; measurement versus assumption=explicit | Unavailable — latency-SLA breach is a deterministic fixture pending exact-ID, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch55-mistral-large-vs-mistral-medium-m3-r4validation failure | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai or declared host; hostB=api.mistral.ai or declared host; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=validation failure; trigger evidence; queue/retry state; tokens carried forward; side-effect class; Large eligibility; attempt cap; route/hold/manual outcome; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-02; measurement versus assumption=explicit | Unavailable — validation failure has no matched, dated bilateral observation. | FAIL CLOSED — manual or probe evidence required |
batch55-mistral-large-vs-mistral-medium-m3-r5quota throttle | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai or declared host; hostB=api.mistral.ai or declared host; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=quota throttle; trigger evidence; queue/retry state; tokens carried forward; side-effect class; Large eligibility; attempt cap; route/hold/manual outcome; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-02; measurement versus assumption=explicit | Unavailable — quota throttle is a deterministic fixture pending exact-ID, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch55-mistral-large-vs-mistral-medium-m3-r6repeated Large failure | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai or declared host; hostB=api.mistral.ai or declared host; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=repeated Large failure; trigger evidence; queue/retry state; tokens carried forward; side-effect class; Large eligibility; attempt cap; route/hold/manual outcome; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-02; measurement versus assumption=explicit | Unavailable — repeated Large failure has no matched, dated bilateral observation. | FAIL CLOSED — manual or probe evidence required |
First-party provenance: Mistral API rate limits; verification date 2026-09-02. Missing or conflicting joins fail closed.
Method and limitations: formulas are deterministic; assumptions are labeled; no missing provider, host, account, region, realm, alias, snapshot, version, weight, artifact, effort, tool, modality, workload, rate-period, timestamp, or result is transferred. Run the mistral-large-vs-mistral-medium Batch 55 scenario →
Batch 69 · exact-pair decision contributions · verified 2026-09-08
Exact pair boundary: Mistral; same provider, tier escalation, $0.50/$1.50 vs $1.50/$7.50 token tariffs, 256K context, EU sovereignty must join. Requested models are mistral-large and mistral-medium. Provider, rate, pricing, and task pages remain fact owners.
European sovereign token pricing and monthly spend matrix
Frozen Batch 69 scenario board. Formula / deterministic rule: monthly_spend = calls * ((in_tokens * rate_in + out_tokens * rate_out) / 1M) Boundary: Owns base token tariff comparisons and European enterprise billing.
| Frozen scenario / field ID | Pair, identity, host, surface, and evidence fields | Result | State |
|---|---|---|---|
batch69-mistral-large-vs-mistral-medium-m1-r120K enterprise customer queries | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai; hostB=api.mistral.ai; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=20K enterprise customer queries; workload; prompt tokens; completion tokens; Mistral Large spend; Mistral Medium spend; budget delta; cost winner; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-08; measurement versus assumption=explicit | Unavailable — 20K enterprise customer queries is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch69-mistral-large-vs-mistral-medium-m1-r250K document extraction requests | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai; hostB=api.mistral.ai; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=50K document extraction requests; workload; prompt tokens; completion tokens; Mistral Large spend; Mistral Medium spend; budget delta; cost winner; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-08; measurement versus assumption=explicit | Unavailable — 50K document extraction requests is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch69-mistral-large-vs-mistral-medium-m1-r3100K multilingual translation calls | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai; hostB=api.mistral.ai; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=100K multilingual translation calls; workload; prompt tokens; completion tokens; Mistral Large spend; Mistral Medium spend; budget delta; cost winner; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-08; measurement versus assumption=explicit | Unavailable — 100K multilingual translation calls is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch69-mistral-large-vs-mistral-medium-m1-r4batch API queue (50% discount) | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai; hostB=api.mistral.ai; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=batch API queue (50% discount); workload; prompt tokens; completion tokens; Mistral Large spend; Mistral Medium spend; budget delta; cost winner; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-08; measurement versus assumption=explicit | Unavailable — batch API queue (50% discount) is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch69-mistral-large-vs-mistral-medium-m1-r5EU-only dedicated instance deployment | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai; hostB=api.mistral.ai; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=EU-only dedicated instance deployment; workload; prompt tokens; completion tokens; Mistral Large spend; Mistral Medium spend; budget delta; cost winner; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-08; measurement versus assumption=explicit | Unavailable — EU-only dedicated instance deployment is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch69-mistral-large-vs-mistral-medium-m1-r6unresolved billing currency | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai; hostB=api.mistral.ai; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=unresolved billing currency; workload; prompt tokens; completion tokens; Mistral Large spend; Mistral Medium spend; budget delta; cost winner; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-08; measurement versus assumption=explicit | Unavailable — unresolved billing currency has no matched, dated bilateral observation. | FAIL CLOSED — manual, probe, or source evidence required |
First-party provenance: Mistral AI model pricing; verification date 2026-09-08. Missing or conflicting joins fail closed.
Two-tier enterprise routing: Medium triage vs Large escalation
Frozen Batch 69 scenario board. Formula / deterministic rule: blended_cost = medium_volume * medium_cost + large_volume * large_cost Boundary: Owns two-tier architectural routing between balanced Medium and flagship Large.
| Frozen scenario / field ID | Pair, identity, host, surface, and evidence fields | Result | State |
|---|---|---|---|
batch69-mistral-large-vs-mistral-medium-m2-r1100% Mistral Medium baseline | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai; hostB=api.mistral.ai; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=100% Mistral Medium baseline; escalation mix; total monthly queries; blended spend; cost savings vs pure Large; quality retention; routing verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-08; measurement versus assumption=explicit | Unavailable — 100% Mistral Medium baseline is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch69-mistral-large-vs-mistral-medium-m2-r290% Medium triage / 10% Large escalation | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai; hostB=api.mistral.ai; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=90% Medium triage / 10% Large escalation; escalation mix; total monthly queries; blended spend; cost savings vs pure Large; quality retention; routing verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-08; measurement versus assumption=explicit | Unavailable — 90% Medium triage / 10% Large escalation is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch69-mistral-large-vs-mistral-medium-m2-r380% Medium triage / 20% Large escalation | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai; hostB=api.mistral.ai; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=80% Medium triage / 20% Large escalation; escalation mix; total monthly queries; blended spend; cost savings vs pure Large; quality retention; routing verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-08; measurement versus assumption=explicit | Unavailable — 80% Medium triage / 20% Large escalation is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch69-mistral-large-vs-mistral-medium-m2-r450% Medium triage / 50% Large escalation | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai; hostB=api.mistral.ai; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=50% Medium triage / 50% Large escalation; escalation mix; total monthly queries; blended spend; cost savings vs pure Large; quality retention; routing verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-08; measurement versus assumption=explicit | Unavailable — 50% Medium triage / 50% Large escalation is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch69-mistral-large-vs-mistral-medium-m2-r5100% direct Mistral Large execution | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai; hostB=api.mistral.ai; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=100% direct Mistral Large execution; escalation mix; total monthly queries; blended spend; cost savings vs pure Large; quality retention; routing verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-08; measurement versus assumption=explicit | Unavailable — 100% direct Mistral Large execution is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch69-mistral-large-vs-mistral-medium-m2-r6unresolved routing confidence threshold | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai; hostB=api.mistral.ai; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=unresolved routing confidence threshold; escalation mix; total monthly queries; blended spend; cost savings vs pure Large; quality retention; routing verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-08; measurement versus assumption=explicit | Unavailable — unresolved routing confidence threshold has no matched, dated bilateral observation. | FAIL CLOSED — manual, probe, or source evidence required |
First-party provenance: Mistral AI model pricing; verification date 2026-09-08. Missing or conflicting joins fail closed.
Data sovereignty and GDPR regulatory compliance audit
Frozen Batch 69 scenario board. Formula / deterministic rule: compliance = (eu_sovereignty_required ? Compliant : Review); jurisdictional constraint Boundary: Owns GDPR data residency and multilingual European language accuracy.
| Frozen scenario / field ID | Pair, identity, host, surface, and evidence fields | Result | State |
|---|---|---|---|
batch69-mistral-large-vs-mistral-medium-m3-r1public web query extraction (Both eligible) | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai; hostB=api.mistral.ai; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=public web query extraction (Both eligible); workload type; EU sovereignty required; Mistral Large compliant; Mistral Medium compliant; regulatory recommendation; compliance verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-08; measurement versus assumption=explicit | Unavailable — public web query extraction (Both eligible) is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch69-mistral-large-vs-mistral-medium-m3-r2internal business intelligence report (Both eligible) | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai; hostB=api.mistral.ai; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=internal business intelligence report (Both eligible); workload type; EU sovereignty required; Mistral Large compliant; Mistral Medium compliant; regulatory recommendation; compliance verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-08; measurement versus assumption=explicit | Unavailable — internal business intelligence report (Both eligible) is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch69-mistral-large-vs-mistral-medium-m3-r3strictly regulated EU financial records (Both compliant) | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai; hostB=api.mistral.ai; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=strictly regulated EU financial records (Both compliant); workload type; EU sovereignty required; Mistral Large compliant; Mistral Medium compliant; regulatory recommendation; compliance verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-08; measurement versus assumption=explicit | Unavailable — strictly regulated EU financial records (Both compliant) is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch69-mistral-large-vs-mistral-medium-m3-r4EU healthcare patient identifiable records (Both compliant) | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai; hostB=api.mistral.ai; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=EU healthcare patient identifiable records (Both compliant); workload type; EU sovereignty required; Mistral Large compliant; Mistral Medium compliant; regulatory recommendation; compliance verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-08; measurement versus assumption=explicit | Unavailable — EU healthcare patient identifiable records (Both compliant) is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch69-mistral-large-vs-mistral-medium-m3-r5cross-border data export restriction trigger | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai; hostB=api.mistral.ai; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=cross-border data export restriction trigger; workload type; EU sovereignty required; Mistral Large compliant; Mistral Medium compliant; regulatory recommendation; compliance verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-08; measurement versus assumption=explicit | Unavailable — cross-border data export restriction trigger is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch69-mistral-large-vs-mistral-medium-m3-r6unresolved data residency jurisdiction | pair=mistral-large vs mistral-medium; provider=Mistral; hostA=api.mistral.ai; hostB=api.mistral.ai; surfaceA=Mistral API; surfaceB=Mistral API; requested/effective IDs=mistral-large,mistral-medium; scenario=unresolved data residency jurisdiction; workload type; EU sovereignty required; Mistral Large compliant; Mistral Medium compliant; regulatory recommendation; compliance verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-08; measurement versus assumption=explicit | Unavailable — unresolved data residency jurisdiction has no matched, dated bilateral observation. | FAIL CLOSED — manual, probe, or source evidence required |
First-party provenance: Mistral AI documentation; verification date 2026-09-08. Missing or conflicting joins fail closed.
Method and limitations: formulas are deterministic; observed and assumed inputs are labeled; no missing provider, host, account, region, realm, alias, snapshot, revision, weight, artifact, control, tool, modality, workload, rate-period, timestamp, or result is transferred. Run the mistral-large-vs-mistral-medium Batch 69 scenario →
How do Mistral Large 3 and Mistral Medium 3 compare on specs?
| Mistral Large 3 | Mistral Medium 3 | |
|---|---|---|
| Price (input) | $0.50/M ✓ | $1.50/M |
| Price (output) | $1.50/M ✓ | $7.50/M |
| Blended price | $0.75/M ✓ | $3.00/M |
| Context window | 256,000 tokens | 256,000 tokens |
| Max output | 32,768 tokens | 32,768 tokens |
| Modalities | text, vision | text, vision |
| Reasoning mode | No | No |
| Released | 2025-12 | 2026-04 |
| Speed | 61 t/s | 92 t/s ✓ |
How much do Mistral Large 3 and Mistral Medium 3 cost at scale?
| Tokens / month | Mistral Large 3 | Mistral Medium 3 | Delta |
|---|---|---|---|
| 1,000,000 | $0.75 | $3.00 | $2.25 (4.0×) |
| 10,000,000 | $7.50 | $30.00 | $22.50 (4.0×) |
| 100,000,000 | $75.00 | $300.00 | $225.00 (4.0×) |
Choose Mistral Large 3 if…
- ✓Mistral’s open-weight multimodal flagship
- ✓Low price relative to other frontier models
- ✓EU-hosted option available
- ✓Strong multimodal reasoning and coding at a lower cost than the big-lab flagships.
Choose Mistral Medium 3 if…
- ✓Frontier-class multimodal performance
- ✓Strong agentic and coding capability
- ✓Vision input included
- ✓Balanced production workloads that need Mistral’s current frontier tier.
Run this exact matchup right now
Send the same prompt to Mistral Large 3 and Mistral Medium 3 side by side and see the outputs yourself.
Try Mistral Large 3 vs Mistral Medium 3 FreeWhat are common questions about Mistral Large 3 and Mistral Medium 3?
Is Mistral Large 3 cheaper than Mistral Medium 3?
Mistral Large 3 is cheaper, at $0.75 per million blended tokens vs $3.00 for Mistral Medium 3.
Which has the bigger context window, Mistral Large 3 or Mistral Medium 3?
Both models support 256,000 tokens of context.
Can Mistral Large 3 replace Mistral Medium 3 for coding?
Both are viable for coding. Mistral’s open-weight multimodal flagship (Mistral Large 3) vs Frontier-class multimodal performance (Mistral Medium 3) — pick based on which strength matters more for your workload.
Which is faster, Mistral Large 3 or Mistral Medium 3?
Mistral Medium 3 is faster: 92 t/s vs 61 t/s, measured on our speed benchmarks.
What related comparisons help choose between Mistral Large 3 and Mistral Medium 3?
Mistral Large 3 pricingMistral Medium 3 pricingvs DeepSeek V4 Provs Claude Sonnet 5Cheap model tests
Pricing verified 2026-08-14. Specs verified 2026-08-14.