Cognition launches SWE-2 Flash, claiming three times the speed and one-fifth the price of SWE-2
Cognition announced SWE-2 Flash on November 4, 2026, and reports that the coding model runs at three times the speed and one-fifth the price of its larger SWE-2 sibling. It is available in Devin Desktop and CLI, with free use through November 30, 2026.
Cognition announced SWE-2 Flash on November 4, 2026, and reports that the coding model runs at three times the speed and one-fifth the price of its larger SWE-2 sibling. The announcement was covered by Unite.AI.
What Cognition announced
SWE-2 Flash is available in Devin Desktop and CLI. Cognition positions it for routine software-agent work rather than uniform benchmark leadership. The intended jobs include:
- Automations
- Code review
- Feature development
- Documentation updates
Free use of the model is offered through November 30, 2026.
Key specifications
- Model: SWE-2 Flash
- Developer: Cognition
- Total parameters: 276 billion
- Speed: three times that of SWE-2, according to Cognition
- Price: one-fifth that of SWE-2, according to Cognition
- Availability: Devin Desktop and CLI
- Free period: through November 30, 2026
The speed and price comparisons are Cognition's own claims relative to the larger SWE-2 model.
What this means
Cognition is framing Flash as a model for high-volume, everyday agent tasks, not as a contender for top benchmark scores. That framing matters for teams running agents on repetitive work such as code review and documentation updates, where speed and cost per task weigh more than peak capability. The free period through November 30, 2026 gives users of Devin Desktop and CLI a window to test Flash on those workloads against SWE-2 before any charges apply. Cognition's three-times-speed and one-fifth-price figures are company claims, so real-world results on a team's own tasks are the measure that counts.
Related Articles
ricursiveAI's 65B Ripple Preview launches on GMICloud: 16K context, $0.20/$0.60 per 1M tokens
Ripple-65B-Preview, a 65B-parameter model from ricursiveAI with native tool calling, is served in BF16 by GMICloud and listed on OpenRouter as of Nov 4, 2026. It has a 16K context window and costs $0.20 per 1M input tokens and $0.60 per 1M output tokens.
Tencent releases Hy4 with 1.0M context and first visual understanding, priced at $0.417 / $1.251 per 1M tokens
Tencent has released the official version of Hy4, with a 1.0M-token context window and visual understanding for the first time. It is listed on OpenRouter at $0.417 input / $1.251 output per 1M tokens, shown as 50% off.
OpenAI's Images 2.6 lets ChatGPT generate animated GIFs with speed and frame editing
ChatGPT can now create animated GIFs from a text prompt or an uploaded photo, with controls for playback speed and individual frames. The feature ships with OpenAI's new Images 2.6 model, which OpenAI says also generates images faster. OpenAI says users now generate over 4 billion images per week across ChatGPT and its API.
Dharma-AI's 2B Field Extractor scores 0.834 on its own benchmark, above Claude Opus 5's 0.816
Dharma-AI released Dharma Field Extractor, a 2-billion-parameter model that turns a document image into typed JSON, either open-ended or matched to a supplied schema. According to Dharma-AI, it scored 0.834 on its 1,638-case benchmark, above every frontier configuration tested, and runs self-hosted on a single L40S.
Comments
Loading...