Skip to main content
AI Choice Engine

Model comparison

Amazon Nova 2 Pro vs SWE-1.5

Amazon against Cognition, compared on context, price, and attributed benchmark results.

Amazon Nova 2 Pro: profile checked October 25, 2026SWE-1.5: profile checked October 25, 2026Individual provider fields may changeEvidence confidence: Insufficient data — see receipts below

Amazon

Amazon Nova 2 Pro

Balanced

vs

Cognition

SWE-1.5

Balanced

AI model capability comparison
SpecificationAmazon Nova 2 ProSWE-1.5
ProviderAmazonCognition
TierBalancedBalanced
Context window1MNot verifiedUnverified
Max outputNot verifiedUnverifiedNot verifiedUnverified
Input / 1M tokens$1.25Not verifiedUnverified
Output / 1M tokens$10Not verifiedUnverified
WeightsClosedNot verifiedUnverified
ParametersNot verifiedUnverifiedHundreds of billions (publisher; exact count unverified)
Reasoning levelslow, medium, highNot verifiedUnverified
Modalitiestext, image, video, audiotext
ReleasedDecember 2, 2025October 29, 2025

Dates beside benchmark scores refer to the linked source’s publication or capture date unless that source explicitly states when it measured the result. A shared URL alone does not establish comparable runs; unverified setups are marked. Prices are USD per million tokens at standard rates, excluding batch and caching discounts. Values we could not confirm from the provider are shown as “Not verified” rather than estimated.

Pricing tiers

Amazon Nova 2 Pro: Bedrock Global Standard $1.25/$10 per MTok; single-region $1.375/$11. Global Flex/Batch $0.625/$5 and regional $0.6875/$5.50. Global Priority $2.1875/$17.50 and regional $2.40625/$19.25. These are native Nova 2.0 Pro on-demand meters in the complete offer published October 16, 2026, with October 1 effective dates; not custom-model, provisioned-throughput or training prices. No Pro cached-input meter is present in this capture. Billing meters do not establish GA or account access.

SWE-1.5: Historical Windsurf model; the current guide does not list the original SWE-1.5 separately. Quota docs mention SWE-1.5 Fast only as an example of speed configurations that can cost more. Neither that example nor launch promotions establish a current route, account entitlement or token tariff for the original model.

BalancedRecord checked October 25, 2026

Amazon Nova 2 Pro

Amazon's closed Nova 2 Pro preview accepts text, images, video and speech and generates text; its launch announcement specifies a one-million-token context.

Best for

  • AWS-native workloads
  • Multimodal
  • Long-context

Watch out

Preview access was announced for Nova Forge customers through their AWS account team. Current account access, a stable API model ID and a hard output cap remain unverified. Launch documentation names low/medium/high thinking; the current detailed thinking guide covers Lite only, so do not transfer its request rules to Pro. The October 1 Artificial Analysis 14-point Pro row is an estimate and is withheld. Omission from a current model list does not prove retirement.

BalancedRecord checked October 25, 2026

SWE-1.5

A prior software-engineering agent model documented in the Windsurf launch and retained here as a historical reference.

Best for

  • Understanding earlier SWE agent references and the 1.6 training lineage

Watch out

Exact context/output caps, stable generic inference API ID, downloadable weights/licence and final model-specific numeric parameter count remain unverified. Authenticated List Models returns account menu UIDs, prices and allowlists; its placeholder examples are not live observations or inference endpoints. Product billing, quota, legacy credits and ACUs must not be conflated. The launch discloses hundreds of billions of parameters without an exact count. Up-to-950-TPS and cross-model comparisons are historical vendor claims involving co-designed inference and harnesses, not current generic API performance.

When the cheaper one wins

One of these rows is missing a verified output price, so this page will not name a cheaper winner. Run the model picker against the actual job instead of guessing.

Run the model picker

Evidence confidence: Insufficient data

How strong and complete the evidence behind this comparison is — not a prediction of which model is better. Hard limitations apply: missing pricing; evidence too sparse.

  • Pricing incomplete — No verified token pricing for SWE-1.5.
  • 1/5 core specs verified on both sides — Not published for at least one side: context window, max output, parameter count, reasoning levels.
  • No directly aligned benchmark setup — No benchmark has matching cited sources, source dates, units, effort and recorded run details on both models.
  • Verified within the last 90 days — Older of the two catalog checks was 1 day ago.
  • Both models carry source citations — Each side has at least two catalog sources on record.

Source receipts

Each catalog figure was checked against the provider or an independent second source on the date shown.

Related comparisons

Diving deeper on one model? Amazon Nova 2 Pro · SWE-1.5

Common questions

Amazon Nova 2 Pro vs SWE-1.5

Answered from the verified figures on this page rather than general guidance.

Should I use Amazon Nova 2 Pro or SWE-1.5?
Both are classified in our balanced tier; this editorial grouping does not imply equal capability. Amazon Nova 2 Pro is a candidate for aws-native workloads; SWE-1.5 is a candidate for understanding earlier swe agent references and the 1.6 training lineage. Compare accuracy, tool reliability, latency and cost on representative tasks.

Next step

Choosing between them

The verified figures that separate this pair, computed from the catalog rather than restated boilerplate.

These two models match on every verified catalog figure — the decision comes down to provider preference, availability, and workload fit rather than published numbers.

Run the model pickerEstimate stack costAll model comparisons

Choosing a harness rather than a model — Cursor, Copilot, Claude Code, Windsurf, or Lovable? Compare agentic harnesses · Latest releases.