Compare Amazon Nova Lite and Nova Pro: Budget vs Capable Multimodal Intelligence
Amazon Nova Lite vs Amazon Nova Pro: which should I use?
Amazon Nova Lite delivers budget efficiency at $0.06/M input and $0.24/M output, while Nova Pro offers balanced multimodal reasoning ($0.80/M input, $3.20/M output) for complex document understanding. Verified 2026-09-07.
Verified 2026-09-07
Where can you find price, speed, and task evidence for Amazon Nova Lite and Amazon Nova Pro?
Which tasks fit Amazon Nova Lite and Amazon Nova Pro?
Best-fit task signals from the published model strengths; this is not a substitute for a controlled benchmark.
| Fact | Amazon Nova Lite | Amazon Nova Pro |
|---|---|---|
| Best fit | Cost-sensitive multimodal tasks on AWS Bedrock. | AWS-native multimodal workloads with a long context requirement. |
| Reasoning mode | Unavailable | Unavailable |
What does a Coding Agent workload cost with Amazon Nova Lite and Amazon Nova Pro?
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 | Amazon Nova Lite | Amazon Nova Pro |
|---|---|---|
| Coding Agent / task | $0.0017 (modeled) (winner) | $0.02 (modeled) |
| Input / output rate | $0.06 / $0.24 per M | $0.80 / $3.20 per M |
How fast are Amazon Nova Lite and Amazon Nova Pro?
Only non-estimated benchmark results are shown as measured.
| Fact | Amazon Nova Lite | Amazon Nova Pro |
|---|---|---|
| Measured throughput | 108 tokens/s (winner) | 64 tokens/s |
| Time to first token | 250 ms | 350 ms |
How compatible are Amazon Nova Lite and Amazon Nova Pro with APIs?
Provider-level wire-format and SDK facts; model-specific parameter differences may still apply.
| Fact | Amazon Nova Lite | Amazon Nova Pro |
|---|---|---|
| OpenAI SDK | Not drop-in | Not drop-in |
| Request shape | Bedrock Converse API — no bearer key; every request is signed with AWS SigV4 using IAM credentials. | Bedrock Converse API — no bearer key; every request is signed with AWS SigV4 using IAM credentials. |
| Streaming | AWS event-stream (ConverseStream), not plain SSE | AWS event-stream (ConverseStream), not plain SSE |
What are the privacy and retention policies for Amazon Nova Lite and Amazon Nova Pro?
Provider policy and residency facts from the verified provider registry. A missing retention policy is not treated as a guarantee.
| Fact | Amazon Nova Lite | Amazon Nova Pro |
|---|---|---|
| Provider says API data trains models | No | No |
| Published retention period | Unavailable | Unavailable |
| Data residency | Selectable AWS region | Selectable AWS region |
How much effort does it take to migrate between Amazon Nova Lite and Amazon Nova Pro?
Direction is from each displayed model to the other. Effort is derived from provider compatibility and published parameter maps.
| Fact | Amazon Nova Lite | Amazon Nova Pro |
|---|---|---|
| Move Amazon Nova Lite → Amazon Nova Pro | drop-in; 0 breaking parameter differences | Target: Amazon Nova Pro |
| Move Amazon Nova Pro → Amazon Nova Lite | Source: Amazon Nova Pro | drop-in; 0 breaking parameter differences |
| Why | Same provider; only the model identifier changes. | Same provider; only the model identifier changes. |
Batch 54 · exact pair cutover and operating-tier contributions. Surface verification: 2026-09-02. These exactly three boards are server-rendered from frozen scenario fixtures; assumptions and unavailable joins are not observations.
Exact pair boundary: Amazon Bedrock; account and region are identity keys; direct IDs, inference profiles, and lifecycle must be joined. Requested models are nova-lite and nova-pro. Provider, rate, pricing, and task pages remain fact owners.
Bedrock model-and-inference-profile admission matrix
Frozen Batch 54 scenario board. Formula / deterministic rule: eligible = account + region + effective ID/profile + modality/tool/schema evidence; unknown region/profile => block Boundary: Exact Nova Lite-versus-Pro Bedrock admission only; Premier, Micro, provider adoption, and arbitrary AWS totals remain out of scope.
| Frozen scenario / field ID | Pair, host, surface, and evidence fields | Result | State |
|---|---|---|---|
batch54-nova-lite-vs-nova-pro-m1-r1direct regional ID | pair=nova-lite vs nova-pro; provider/host/surface=bedrock-runtime.{region}.amazonaws.com / bedrock-runtime.{region}.amazonaws.com; Bedrock Converse API / Bedrock Converse API; requested IDs=nova-lite,nova-pro; effective IDs=nova-lite,nova-pro; account=required; region=us-east-1; direct IDs=exact; requested/effective Lite and Pro IDs=required; modality/schema=declared; availability evidence=dated; evidence=2026-09-02 | Admit a tier only when the exact account and region return the requested effective ID. | CONDITIONAL — regional admission |
batch54-nova-lite-vs-nova-pro-m1-r2cross-region profile | pair=nova-lite vs nova-pro; provider/host/surface=bedrock-runtime.{region}.amazonaws.com / bedrock-runtime.{region}.amazonaws.com; Bedrock Converse API / Bedrock Converse API; requested IDs=nova-lite,nova-pro; effective IDs=nova-lite,nova-pro; account=required; region=source+target; profile=cross-region; inference profile ID=required; source/target region=required; effective tier=required; data path=declared; evidence=2026-09-02 | Unavailable — cross-region profile and effective tier are not joined | UNAVAILABLE — profile admission |
batch54-nova-lite-vs-nova-pro-m1-r3unsupported region | pair=nova-lite vs nova-pro; provider/host/surface=bedrock-runtime.{region}.amazonaws.com / bedrock-runtime.{region}.amazonaws.com; Bedrock Converse API / Bedrock Converse API; requested IDs=nova-lite,nova-pro; effective IDs=nova-lite,nova-pro; account=required; region=unsupported; requested model ID=known; effective ID=none; availability evidence=dated; fallback region=not assumed; evidence=2026-09-02 | Do not transfer availability from another region; hold until an admitted region is selected. | BLOCKED — region unsupported |
batch54-nova-lite-vs-nova-pro-m1-r4provisioned candidate | pair=nova-lite vs nova-pro; provider/host/surface=bedrock-runtime.{region}.amazonaws.com / bedrock-runtime.{region}.amazonaws.com; Bedrock Converse API / Bedrock Converse API; requested IDs=nova-lite,nova-pro; effective IDs=nova-lite,nova-pro; account=required; region=exact; deployment=provisioned; provisioned model/profile ID=required; capacity/quota=required; effective tier=required; rate period=dated; evidence=2026-09-02 | Unavailable — provisioned capacity and quota evidence are missing | UNAVAILABLE — provisioned admission |
batch54-nova-lite-vs-nova-pro-m1-r5gateway alias | pair=nova-lite vs nova-pro; provider/host/surface=shared gateway / shared gateway; gateway / gateway; requested IDs=nova-lite,nova-pro; effective IDs=nova-lite,nova-pro; account/region=unknown; requested=latest; effective Lite/Pro ID/profile=unknown; modality/tool/schema=unknown; provider=unknown; evidence=2026-09-02 | Reject a gateway alias; Bedrock admission is account- and region-qualified. | BLOCKED — gateway identity |
batch54-nova-lite-vs-nova-pro-m1-r6unknown account | pair=nova-lite vs nova-pro; provider/host/surface=bedrock-runtime.{region}.amazonaws.com / bedrock-runtime.{region}.amazonaws.com; Bedrock Converse API / Bedrock Converse API; requested IDs=nova-lite,nova-pro; effective IDs=nova-lite,nova-pro; account=unknown; region=known; IDs=known; effective profile=unknown; entitlement/quota=unknown; availability evidence=missing; evidence=2026-09-02 | No cross-account availability assumption; require the target account entitlement and quota join. | BLOCKED — account unknown |
First-party provenance: Amazon Nova model documentation; verification date 2026-09-02. Missing or conflicting joins fail closed.
Multimodal task evidence receipt
Frozen Batch 54 scenario board. Formula / deterministic rule: supported = exact tier/profile + artifact hash + matched prompt/config/rubric/denominator; otherwise Unavailable Boundary: This receipt owns exact Nova Lite/Pro task evidence; broad image and document recommendations remain task-page owners.
| Frozen scenario / field ID | Pair, host, surface, and evidence fields | Result | State |
|---|---|---|---|
batch54-nova-lite-vs-nova-pro-m2-r1text summary | pair=nova-lite vs nova-pro; provider/host/surface=bedrock-runtime.{region}.amazonaws.com / bedrock-runtime.{region}.amazonaws.com; Bedrock Converse API / Bedrock Converse API; requested IDs=nova-lite,nova-pro; effective IDs=nova-lite,nova-pro; region=exact; profile=exact; artifact=text-001; prompt/config hash=paired; result hashes=required; rubric/denominator=required; latency/tokens=required; evidence=2026-09-02 | Unavailable — paired summary acceptance and telemetry are missing | UNTESTED — text evidence |
batch54-nova-lite-vs-nova-pro-m2-r2scanned-document extraction | pair=nova-lite vs nova-pro; provider/host/surface=bedrock-runtime.{region}.amazonaws.com / bedrock-runtime.{region}.amazonaws.com; Bedrock Converse API / Bedrock Converse API; requested IDs=nova-lite,nova-pro; effective IDs=nova-lite,nova-pro; region=exact; profile=exact; artifact=doc-002; document hash=doc-002; OCR/layout state=required; schema hash=required; field rubric/denominator=required; evidence=2026-09-02 | Unavailable — matched document field results are absent | UNAVAILABLE — document evidence |
batch54-nova-lite-vs-nova-pro-m2-r3image comparison | pair=nova-lite vs nova-pro; provider/host/surface=bedrock-runtime.{region}.amazonaws.com / bedrock-runtime.{region}.amazonaws.com; Bedrock Converse API / Bedrock Converse API; requested IDs=nova-lite,nova-pro; effective IDs=nova-lite,nova-pro; region=exact; profile=exact; artifact=image-003; image hashes/order=required; prompt/config=paired; rubric=required; result hash=missing; evidence=2026-09-02 | Unavailable — paired image result and latency are not joined | UNAVAILABLE — image evidence |
batch54-nova-lite-vs-nova-pro-m2-r4short-video analysis | pair=nova-lite vs nova-pro; provider/host/surface=bedrock-runtime.{region}.amazonaws.com / bedrock-runtime.{region}.amazonaws.com; Bedrock Converse API / Bedrock Converse API; requested IDs=nova-lite,nova-pro; effective IDs=nova-lite,nova-pro; region=exact; profile=exact; artifact=video-004; video hash=video-004; frame/audio units=required; output schema=required; result/latency=Untested; evidence=2026-09-02 | Unavailable — video-unit accounting and matched result are missing | UNAVAILABLE — video evidence |
batch54-nova-lite-vs-nova-pro-m2-r5tool-augmented lookup | pair=nova-lite vs nova-pro; provider/host/surface=bedrock-runtime.{region}.amazonaws.com / bedrock-runtime.{region}.amazonaws.com; Bedrock Converse API / Bedrock Converse API; requested IDs=nova-lite,nova-pro; effective IDs=nova-lite,nova-pro; region=exact; profile=exact; artifact=tool-005; tool schema/call IDs=required; side_effect=read-only; citations=required; prompt/config hash=paired; evidence=2026-09-02 | Unavailable — tool trace and source support are not paired | UNAVAILABLE — tool evidence |
batch54-nova-lite-vs-nova-pro-m2-r6missing-Pro-or-Lite-result | pair=nova-lite vs nova-pro; provider/host/surface=bedrock-runtime.{region}.amazonaws.com / bedrock-runtime.{region}.amazonaws.com; Bedrock Converse API / Bedrock Converse API; requested IDs=nova-lite,nova-pro; effective IDs=nova-lite,nova-pro; region=exact; profile=exact; artifact=missing-006; one tier result=missing; prompt/config/rubric=partial; latency/tokens=Untested; denominator=missing; evidence=2026-09-02 | A missing side cannot establish Pro superiority or Lite eligibility. | UNAVAILABLE — bilateral result required |
First-party provenance: Amazon Nova model documentation; verification date 2026-09-02. Missing or conflicting joins fail closed.
Lite-first/Pro-escalation operations board
Frozen Batch 54 scenario board. Formula / deterministic rule: escalate = admitted Pro + observed trigger + bounded retry; promote = floor + quota/failure policy + side-effect containment Boundary: Only Nova Lite/Pro Bedrock escalation is owned; rate links are dated inputs, not an arbitrary cost verdict.
| Frozen scenario / field ID | Pair, host, surface, and evidence fields | Result | State |
|---|---|---|---|
batch54-nova-lite-vs-nova-pro-m3-r1Lite-only | pair=nova-lite vs nova-pro; provider/host/surface=bedrock-runtime.{region}.amazonaws.com / bedrock-runtime.{region}.amazonaws.com; Bedrock Converse API / Bedrock Converse API; requested IDs=nova-lite,nova-pro; effective IDs=nova-lite,nova-pro; route=lite-only; scenario=ops-001; region=exact; Lite admission=required; acceptance floor=declared; Pro not called; attempt cap=1; fallback=human; evidence=2026-09-02 | Keep Lite-only only after its local acceptance floor clears. Measurement state: labeled assumption only. | ASSUMPTION ONLY — local evidence required |
batch54-nova-lite-vs-nova-pro-m3-r2Pro-only | pair=nova-lite vs nova-pro; provider/host/surface=bedrock-runtime.{region}.amazonaws.com / bedrock-runtime.{region}.amazonaws.com; Bedrock Converse API / Bedrock Converse API; requested IDs=nova-lite,nova-pro; effective IDs=nova-lite,nova-pro; route=pro-only; scenario=ops-002; region=exact; consequence=critical; Pro profile=admitted; expert rubric=required; quota/failure=dated; evidence=2026-09-02 | Use Pro-only only under a declared consequence rule; no tier-name inference. | CONDITIONAL — consequence rule |
batch54-nova-lite-vs-nova-pro-m3-r3confidence trigger | pair=nova-lite vs nova-pro; provider/host/surface=bedrock-runtime.{region}.amazonaws.com / bedrock-runtime.{region}.amazonaws.com; Bedrock Converse API / Bedrock Converse API; requested IDs=nova-lite,nova-pro; effective IDs=nova-lite,nova-pro; route=lite→pro; scenario=ops-003; region=exact; confidence classifier hash=required; trigger evidence=required; tokens/artifacts carried forward=declared; attempt cap=2; evidence=2026-09-02 | Unavailable — trigger calibration and Pro recovery are not observed | UNAVAILABLE — confidence trigger |
batch54-nova-lite-vs-nova-pro-m3-r4schema-failure trigger | pair=nova-lite vs nova-pro; provider/host/surface=bedrock-runtime.{region}.amazonaws.com / bedrock-runtime.{region}.amazonaws.com; Bedrock Converse API / Bedrock Converse API; requested IDs=nova-lite,nova-pro; effective IDs=nova-lite,nova-pro; route=lite→pro; scenario=ops-004; region=exact; schema hash=paired; validator failure=required; retry policy=declared; side_effect=read-only; rate link=dated; evidence=2026-09-02 | Unavailable — schema failure rate and escalation value are missing | UNAVAILABLE — schema trigger |
batch54-nova-lite-vs-nova-pro-m3-r5regional failover | pair=nova-lite vs nova-pro; provider/host/surface=bedrock-runtime.{region}.amazonaws.com / bedrock-runtime.{region}.amazonaws.com; Bedrock Converse API / Bedrock Converse API; requested IDs=nova-lite,nova-pro; effective IDs=nova-lite,nova-pro; route=Lite/Pro regional failover; scenario=ops-005; source/target regions=required; profile IDs=required; quota/failure class=required; replay=required; evidence=2026-09-02 | Fail over only to an admitted exact profile and region; do not assume cross-region equivalence. | CONDITIONAL — regional proof |
batch54-nova-lite-vs-nova-pro-m3-r6repeated failure | pair=nova-lite vs nova-pro; provider/host/surface=bedrock-runtime.{region}.amazonaws.com / bedrock-runtime.{region}.amazonaws.com; Bedrock Converse API / Bedrock Converse API; requested IDs=nova-lite,nova-pro; effective IDs=nova-lite,nova-pro; route=lite→pro→human; scenario=ops-006; region=exact; attempt cap=2; failure classes=required; idempotency/side_effect=required; terminal reviewer=required; evidence=2026-09-02 | Stop after two attempts, settle effects, and send the case to human review or rollback. | BLOCKED — terminal human action |
First-party provenance: Amazon Bedrock quotas; 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, cluster, alias, snapshot, version, artifact, effort, tool, modality, benchmark, workload, rate-period, timestamp, or result is transferred. Run the nova-lite-vs-nova-pro Batch 54 scenario →
Batch 64 · exact-pair decision contributions · verified 2026-09-07
Exact pair boundary: Amazon; same provider, tier escalation, multimodal OCR, and Bedrock provisioned throughput must join. Requested models are nova-lite and nova-pro. Provider, rate, pricing, and task pages remain fact owners.
Tier escalation and portfolio routing financial model
Frozen Batch 64 scenario board. Formula / deterministic rule: blended_spend = lite_volume * lite_cost + pro_volume * pro_cost Boundary: Owns two-tier architectural routing between Nova Lite and Nova Pro.
| Frozen scenario / field ID | Pair, identity, host, surface, and evidence fields | Result | State |
|---|---|---|---|
batch64-nova-lite-vs-nova-pro-m1-r1100% Nova Lite baseline ($0.06/$0.24) | pair=nova-lite vs nova-pro; provider=Amazon; hostA=bedrock-runtime.us-east-1.amazonaws.com; hostB=bedrock-runtime.us-east-1.amazonaws.com; surfaceA=AWS Bedrock Converse API; surfaceB=AWS Bedrock Converse API; requested/effective IDs=nova-lite,nova-pro; scenario=100% Nova Lite baseline ($0.06/$0.24); escalation mix; total monthly queries; blended spend; savings vs pure Pro; accuracy retention; routing verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — 100% Nova Lite baseline ($0.06/$0.24) is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch64-nova-lite-vs-nova-pro-m1-r290% Lite / 10% Pro escalation | pair=nova-lite vs nova-pro; provider=Amazon; hostA=bedrock-runtime.us-east-1.amazonaws.com; hostB=bedrock-runtime.us-east-1.amazonaws.com; surfaceA=AWS Bedrock Converse API; surfaceB=AWS Bedrock Converse API; requested/effective IDs=nova-lite,nova-pro; scenario=90% Lite / 10% Pro escalation; escalation mix; total monthly queries; blended spend; savings vs pure Pro; accuracy retention; routing verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — 90% Lite / 10% Pro escalation is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch64-nova-lite-vs-nova-pro-m1-r375% Lite / 25% Pro escalation | pair=nova-lite vs nova-pro; provider=Amazon; hostA=bedrock-runtime.us-east-1.amazonaws.com; hostB=bedrock-runtime.us-east-1.amazonaws.com; surfaceA=AWS Bedrock Converse API; surfaceB=AWS Bedrock Converse API; requested/effective IDs=nova-lite,nova-pro; scenario=75% Lite / 25% Pro escalation; escalation mix; total monthly queries; blended spend; savings vs pure Pro; accuracy retention; routing verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — 75% Lite / 25% Pro escalation is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch64-nova-lite-vs-nova-pro-m1-r450% Lite / 50% Pro escalation | pair=nova-lite vs nova-pro; provider=Amazon; hostA=bedrock-runtime.us-east-1.amazonaws.com; hostB=bedrock-runtime.us-east-1.amazonaws.com; surfaceA=AWS Bedrock Converse API; surfaceB=AWS Bedrock Converse API; requested/effective IDs=nova-lite,nova-pro; scenario=50% Lite / 50% Pro escalation; escalation mix; total monthly queries; blended spend; savings vs pure Pro; accuracy retention; routing verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — 50% Lite / 50% Pro escalation is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch64-nova-lite-vs-nova-pro-m1-r5100% Nova Pro comparison ($0.80/$3.20) | pair=nova-lite vs nova-pro; provider=Amazon; hostA=bedrock-runtime.us-east-1.amazonaws.com; hostB=bedrock-runtime.us-east-1.amazonaws.com; surfaceA=AWS Bedrock Converse API; surfaceB=AWS Bedrock Converse API; requested/effective IDs=nova-lite,nova-pro; scenario=100% Nova Pro comparison ($0.80/$3.20); escalation mix; total monthly queries; blended spend; savings vs pure Pro; accuracy retention; routing verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — 100% Nova Pro comparison ($0.80/$3.20) is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch64-nova-lite-vs-nova-pro-m1-r6unresolved confidence threshold | pair=nova-lite vs nova-pro; provider=Amazon; hostA=bedrock-runtime.us-east-1.amazonaws.com; hostB=bedrock-runtime.us-east-1.amazonaws.com; surfaceA=AWS Bedrock Converse API; surfaceB=AWS Bedrock Converse API; requested/effective IDs=nova-lite,nova-pro; scenario=unresolved confidence threshold; escalation mix; total monthly queries; blended spend; savings vs pure Pro; accuracy retention; routing verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — unresolved confidence threshold has no matched, dated bilateral observation. | FAIL CLOSED — manual, probe, or source evidence required |
First-party provenance: AWS Bedrock Nova pricing; verification date 2026-09-07. Missing or conflicting joins fail closed.
Multimodal vision and document OCR extraction accuracy gate
Frozen Batch 64 scenario board. Formula / deterministic rule: roi = document_value_extracted - ((in_tokens * rate_in + out_tokens * rate_out) / 1M) Boundary: Owns OCR and multimodal document analysis return on investment.
| Frozen scenario / field ID | Pair, identity, host, surface, and evidence fields | Result | State |
|---|---|---|---|
batch64-nova-lite-vs-nova-pro-m2-r1simple receipt text extraction (Lite optimal) | pair=nova-lite vs nova-pro; provider=Amazon; hostA=bedrock-runtime.us-east-1.amazonaws.com; hostB=bedrock-runtime.us-east-1.amazonaws.com; surfaceA=AWS Bedrock Converse API; surfaceB=AWS Bedrock Converse API; requested/effective IDs=nova-lite,nova-pro; scenario=simple receipt text extraction (Lite optimal); document type; page count; Lite accuracy; Pro accuracy; recommended model; economic recommendation; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — simple receipt text extraction (Lite optimal) is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch64-nova-lite-vs-nova-pro-m2-r2invoice key-value parsing (Lite optimal) | pair=nova-lite vs nova-pro; provider=Amazon; hostA=bedrock-runtime.us-east-1.amazonaws.com; hostB=bedrock-runtime.us-east-1.amazonaws.com; surfaceA=AWS Bedrock Converse API; surfaceB=AWS Bedrock Converse API; requested/effective IDs=nova-lite,nova-pro; scenario=invoice key-value parsing (Lite optimal); document type; page count; Lite accuracy; Pro accuracy; recommended model; economic recommendation; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — invoice key-value parsing (Lite optimal) is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch64-nova-lite-vs-nova-pro-m2-r3complex multi-page legal contract (Pro optimal) | pair=nova-lite vs nova-pro; provider=Amazon; hostA=bedrock-runtime.us-east-1.amazonaws.com; hostB=bedrock-runtime.us-east-1.amazonaws.com; surfaceA=AWS Bedrock Converse API; surfaceB=AWS Bedrock Converse API; requested/effective IDs=nova-lite,nova-pro; scenario=complex multi-page legal contract (Pro optimal); document type; page count; Lite accuracy; Pro accuracy; recommended model; economic recommendation; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — complex multi-page legal contract (Pro optimal) is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch64-nova-lite-vs-nova-pro-m2-r4technical schematic diagram OCR (Pro optimal) | pair=nova-lite vs nova-pro; provider=Amazon; hostA=bedrock-runtime.us-east-1.amazonaws.com; hostB=bedrock-runtime.us-east-1.amazonaws.com; surfaceA=AWS Bedrock Converse API; surfaceB=AWS Bedrock Converse API; requested/effective IDs=nova-lite,nova-pro; scenario=technical schematic diagram OCR (Pro optimal); document type; page count; Lite accuracy; Pro accuracy; recommended model; economic recommendation; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — technical schematic diagram OCR (Pro optimal) is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch64-nova-lite-vs-nova-pro-m2-r5resolution downsampling penalty | pair=nova-lite vs nova-pro; provider=Amazon; hostA=bedrock-runtime.us-east-1.amazonaws.com; hostB=bedrock-runtime.us-east-1.amazonaws.com; surfaceA=AWS Bedrock Converse API; surfaceB=AWS Bedrock Converse API; requested/effective IDs=nova-lite,nova-pro; scenario=resolution downsampling penalty; document type; page count; Lite accuracy; Pro accuracy; recommended model; economic recommendation; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — resolution downsampling penalty is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch64-nova-lite-vs-nova-pro-m2-r6unsupported image MIME type | pair=nova-lite vs nova-pro; provider=Amazon; hostA=bedrock-runtime.us-east-1.amazonaws.com; hostB=bedrock-runtime.us-east-1.amazonaws.com; surfaceA=AWS Bedrock Converse API; surfaceB=AWS Bedrock Converse API; requested/effective IDs=nova-lite,nova-pro; scenario=unsupported image MIME type; document type; page count; Lite accuracy; Pro accuracy; recommended model; economic recommendation; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — unsupported image MIME type has no matched, dated bilateral observation. | FAIL CLOSED — manual, probe, or source evidence required |
First-party provenance: Amazon Nova documentation; verification date 2026-09-07. Missing or conflicting joins fail closed.
AWS Bedrock provisioned throughput break-even calculator
Frozen Batch 64 scenario board. Formula / deterministic rule: on_demand = calls * ((in * rate_in + out * rate_out) / 1M); provisioned = units * hourly_rate * 730 Boundary: Owns Bedrock financial commitment and capacity planning.
| Frozen scenario / field ID | Pair, identity, host, surface, and evidence fields | Result | State |
|---|---|---|---|
batch64-nova-lite-vs-nova-pro-m3-r1low intermittent traffic (on-demand optimal) | pair=nova-lite vs nova-pro; provider=Amazon; hostA=bedrock-runtime.us-east-1.amazonaws.com; hostB=bedrock-runtime.us-east-1.amazonaws.com; surfaceA=AWS Bedrock Converse API; surfaceB=AWS Bedrock Converse API; requested/effective IDs=nova-lite,nova-pro; scenario=low intermittent traffic (on-demand optimal); monthly call volume; on-demand cost; provisioned cost; utilization break-even point; purchase recommendation; verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — low intermittent traffic (on-demand optimal) is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch64-nova-lite-vs-nova-pro-m3-r2500K steady monthly calls | pair=nova-lite vs nova-pro; provider=Amazon; hostA=bedrock-runtime.us-east-1.amazonaws.com; hostB=bedrock-runtime.us-east-1.amazonaws.com; surfaceA=AWS Bedrock Converse API; surfaceB=AWS Bedrock Converse API; requested/effective IDs=nova-lite,nova-pro; scenario=500K steady monthly calls; monthly call volume; on-demand cost; provisioned cost; utilization break-even point; purchase recommendation; verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — 500K steady monthly calls is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch64-nova-lite-vs-nova-pro-m3-r32M high-volume enterprise traffic | pair=nova-lite vs nova-pro; provider=Amazon; hostA=bedrock-runtime.us-east-1.amazonaws.com; hostB=bedrock-runtime.us-east-1.amazonaws.com; surfaceA=AWS Bedrock Converse API; surfaceB=AWS Bedrock Converse API; requested/effective IDs=nova-lite,nova-pro; scenario=2M high-volume enterprise traffic; monthly call volume; on-demand cost; provisioned cost; utilization break-even point; purchase recommendation; verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — 2M high-volume enterprise traffic is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch64-nova-lite-vs-nova-pro-m3-r410M constant production load | pair=nova-lite vs nova-pro; provider=Amazon; hostA=bedrock-runtime.us-east-1.amazonaws.com; hostB=bedrock-runtime.us-east-1.amazonaws.com; surfaceA=AWS Bedrock Converse API; surfaceB=AWS Bedrock Converse API; requested/effective IDs=nova-lite,nova-pro; scenario=10M constant production load; monthly call volume; on-demand cost; provisioned cost; utilization break-even point; purchase recommendation; verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — 10M constant production load is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch64-nova-lite-vs-nova-pro-m3-r5uncommitted provisioned hour idle | pair=nova-lite vs nova-pro; provider=Amazon; hostA=bedrock-runtime.us-east-1.amazonaws.com; hostB=bedrock-runtime.us-east-1.amazonaws.com; surfaceA=AWS Bedrock Converse API; surfaceB=AWS Bedrock Converse API; requested/effective IDs=nova-lite,nova-pro; scenario=uncommitted provisioned hour idle; monthly call volume; on-demand cost; provisioned cost; utilization break-even point; purchase recommendation; verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — uncommitted provisioned hour idle is a frozen fixture pending exact identity, configuration, and denominator joins. | UNTESTED — assumption cannot establish a verdict |
batch64-nova-lite-vs-nova-pro-m3-r6unsupported model unit tier | pair=nova-lite vs nova-pro; provider=Amazon; hostA=bedrock-runtime.us-east-1.amazonaws.com; hostB=bedrock-runtime.us-east-1.amazonaws.com; surfaceA=AWS Bedrock Converse API; surfaceB=AWS Bedrock Converse API; requested/effective IDs=nova-lite,nova-pro; scenario=unsupported model unit tier; monthly call volume; on-demand cost; provisioned cost; utilization break-even point; purchase recommendation; verdict; prompt/config/input/output/result/cache/checkpoint/artifact hashes=required; evidence=2026-09-07; measurement versus assumption=explicit | Unavailable — unsupported model unit tier has no matched, dated bilateral observation. | FAIL CLOSED — manual, probe, or source evidence required |
First-party provenance: AWS Bedrock Nova pricing; verification date 2026-09-07. 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 nova-lite-vs-nova-pro Batch 64 scenario →
How do Amazon Nova Lite and Amazon Nova Pro compare on specs?
| Amazon Nova Lite | Amazon Nova Pro | |
|---|---|---|
| Price (input) | $0.06/M ✓ | $0.80/M |
| Price (output) | $0.24/M ✓ | $3.20/M |
| Blended price | $0.11/M ✓ | $1.40/M |
| Context window | 300,000 tokens | 300,000 tokens |
| Max output | 16,384 tokens | 32,768 tokens ✓ |
| Modalities | text, vision, audio | text, vision, audio |
| Reasoning mode | No | No |
| Released | 2024-12 | 2024-12 |
| Speed | 108 t/s ✓ | 64 t/s |
How much do Amazon Nova Lite and Amazon Nova Pro cost at scale?
| Tokens / month | Amazon Nova Lite | Amazon Nova Pro | Delta |
|---|---|---|---|
| 1,000,000 | $0.11 | $1.40 | $1.30 (13.3×) |
| 10,000,000 | $1.05 | $14.00 | $12.95 (13.3×) |
| 100,000,000 | $10.50 | $140.00 | $129.50 (13.3×) |
Choose Amazon Nova Lite if…
- ✓Budget-tier accuracy on par with other Flash/Mini models
- ✓Lower output cost than Nova Pro
- ✓Multimodal input included
- ✓Cost-sensitive multimodal tasks on AWS Bedrock.
Choose Amazon Nova Pro if…
- ✓Capable multimodal mid-tier model
- ✓Long context for a Bedrock-native model
- ✓Native audio and video input
- ✓AWS-native multimodal workloads with a long context requirement.
Run this exact matchup right now
Send the same prompt to Amazon Nova Lite and Amazon Nova Pro side by side and see the outputs yourself.
Try Amazon Nova Lite vs Amazon Nova Pro FreeWhat are common questions about Amazon Nova Lite and Amazon Nova Pro?
Is Amazon Nova Lite cheaper than Amazon Nova Pro?
Amazon Nova Lite is cheaper, at $0.11 per million blended tokens vs $1.40 for Amazon Nova Pro.
Which has the bigger context window, Amazon Nova Lite or Amazon Nova Pro?
Both models support 300,000 tokens of context.
Can Amazon Nova Lite replace Amazon Nova Pro for coding?
Both are viable for coding. Budget-tier accuracy on par with other Flash/Mini models (Amazon Nova Lite) vs Capable multimodal mid-tier model (Amazon Nova Pro) — pick based on which strength matters more for your workload.
Which is faster, Amazon Nova Lite or Amazon Nova Pro?
Amazon Nova Lite is faster: 108 t/s vs 64 t/s, measured on our speed benchmarks.
What related comparisons help choose between Amazon Nova Lite and Amazon Nova Pro?
Pricing verified 2026-06-14. Specs verified 2026-08-14.