Model comparison
Base 1 vs SWE-1.5
Base44 against Cognition, compared on context, price, and attributed benchmark results.
Base 1: profile checked October 25, 2026SWE-1.5: profile checked October 25, 2026Individual provider fields may changeEvidence confidence: Insufficient data — see receipts below
Base44
Base 1
Balanced
Cognition
SWE-1.5
Balanced
| Specification | Base 1 | SWE-1.5 |
|---|---|---|
| Provider | ||
| Provider | Base44 | Cognition |
| Tier | ||
| Tier | Balanced | Balanced |
| Context window | ||
| Context window | Not verifiedUnverified | Not verifiedUnverified |
| Max output | ||
| Max output | Not verifiedUnverified | Not verifiedUnverified |
| Input / 1M tokens | ||
| Input / 1M tokens | Not verifiedUnverified | Not verifiedUnverified |
| Output / 1M tokens | ||
| Output / 1M tokens | Not verifiedUnverified | Not verifiedUnverified |
| Weights | ||
| Weights | Not verifiedUnverified | Not verifiedUnverified |
| Parameters | ||
| Parameters | Not verifiedUnverified | Hundreds of billions (publisher; exact count unverified) |
| Reasoning levels | ||
| Reasoning levels | Not verifiedUnverified | Not verifiedUnverified |
| Modalities | ||
| Modalities | text | text |
| Released | ||
| Released | June 29, 2026 | October 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
Base 1: Uses Base44 message credits. Model-specific cost is shown under Credits Used per message; no direct token-rate API price is verified.
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.
Base 1
Base44's first in-house model is offered in its app-building chat, with the publisher describing it as trained on platform building patterns.
Best for
- Evaluating app-building prompts within Base44
- Comparing builder output in an eligible Base44 account
Watch out
The current founder article explicitly says Published June 29, 2026 and 'Today, we announced' Base 1; this establishes the announcement date, not a universal account rollout date. No direct inference endpoint, stable API ID, downloadable checkpoint/licence, exact context/output, parameter count or comparable numeric benchmark is verified. The current manual-selection guide requires Builder or higher and explicitly excludes Base 1 from the effort selector. Per-message Credits Used is authoritative; in-app agent integration-credit prices are a different billing surface. The historical product changelog was not successfully re-fetched, so no broader all-plan availability is asserted.
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 pickerEvidence 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 Base 1 and 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.
- Base 1: Product model access and modes (accessed 2026-10-25)
- Base 1: Publisher model announcement (accessed 2026-10-25)
- Base 1: Product rollout changelog (accessed 2026-10-02)
- Base 1: Credit billing (accessed 2026-10-25)
- SWE-1.5: Current scoped model integration, product billing, entitlement or historical launch evidence (accessed 2026-10-25)
- SWE-1.5: Current scoped model integration, product billing, entitlement or historical launch evidence (accessed 2026-10-25)
- SWE-1.5: Current scoped model integration, product billing, entitlement or historical launch evidence (accessed 2026-10-25)
- SWE-1.5: Current scoped model integration, product billing, entitlement or historical launch evidence (accessed 2026-10-25)
Related comparisons
Common questions
Base 1 vs SWE-1.5
Answered from the verified figures on this page rather than general guidance.
Should I use Base 1 or SWE-1.5?
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.
Choosing a harness rather than a model — Cursor, Copilot, Claude Code, Windsurf, or Lovable? Compare agentic harnesses · Latest releases.