Cognition shipped SWE-1.7 on July 8, 2026 — the most capable model they’ve trained, available today in Devin at 1,000 tokens per second via Cerebras. The headline is benchmark performance at a fraction of frontier cost. But the more interesting story is how they got there: reinforcement learning on top of a model that was already heavily RL-trained. That breaks with the conventional assumption that RL has a ceiling.
Here’s what builders need to know.
Where SWE-1.7 Sits in the July 2026 Model Landscape
The benchmark that matters for direct comparison is Terminal-Bench 2.1 — the same one we’ve used to position Terra, Grok 4.5, and Sonnet 5:
| Model | Terminal-Bench 2.1 | Provider |
|---|---|---|
| GPT-5.6 Terra | 84.3% | OpenAI |
| Grok 4.5 | 83.3% | xAI |
| SWE-1.7 | 81.5% | Cognition |
| Claude Sonnet 5 | 80.4% | Anthropic |
SWE-1.7 slots cleanly between Grok 4.5 and Sonnet 5 — above a frontier Anthropic model, behind a frontier OpenAI and xAI model. Not bad for a model positioned as a cost alternative.
Cognition’s own FrontierCode 1.1 benchmark gives a fuller picture of agentic coding specifically:
| Model | FrontierCode 1.1 |
|---|---|
| Claude Opus 4.8 | 46.5% |
| GPT-5.5 | 43.0% |
| SWE-1.7 | 42.3% |
| Kimi K2.7 (base) | 30.1% |
| Composer 2.5 | 25.6% |
SWE-1.7 sits 0.7 points behind GPT-5.5 and 4.2 points behind Opus 4.8. More strikingly, it’s 12.2 points above its own starting point — the Kimi K2.7 base model it was trained on.
It also scores 77.8% on SWE-Bench Multilingual, Cognition’s cross-language coding benchmark.
The RL-on-RL Breakthrough
This is the technical claim worth taking seriously.
The conventional wisdom has been that once a base model is extensively RL post-trained, you approach a ceiling. Additional RL training on an already-RL-optimized model yields diminishing returns, partly because the model’s entropy collapses — the policy becomes too deterministic and stops exploring new strategies.
Kimi K2.7 (Moonshot AI’s 1-trillion-parameter MoE model with 32B active parameters, released June 2026) had already undergone extensive RL post-training before Cognition touched it. That makes SWE-1.7 a direct test of whether additional RL training stacks.
The result: a 12.2-point jump on FrontierCode 1.1 over the base model. Cognition’s interpretation: “RL can push capabilities much further than previously believed.”
How They Avoided Entropy Collapse
The key technical innovation is entropy preservation using top-p sampling with sampling distribution replay. During RL training, as the model improves, it tends to become overconfident — repeatedly choosing the same high-probability actions and losing the diversity that drives further learning. Cognition’s approach replays the sampling distribution from earlier in training, keeping policy entropy high enough to keep learning.
Self-Compaction for Long Agentic Tasks
Agentic coding tasks can run for hours. SWE-1.7’s training rollouts ran up to six hours — far beyond what fits in a raw context window. Cognition solved this with self-compaction: the model periodically summarizes its own working state and picks up from that summary, using alternating length penalties to drive efficient compression. This means SWE-1.7 can maintain coherent task progress across very long horizons without hitting context limits.
Multi-Cluster Training Infrastructure
To run RL at this scale, Cognition distributed training across compute clusters on three continents with fault tolerance built in. RL training is notoriously brittle to node failures — a single dead node typically kills a run. Their multi-cluster setup keeps training resilient across hardware failures.
Data Curation
An automated pipeline filters out tasks with low learning signal and detects reward exploitation through execution testing. This prevents the model from gaming its training rewards rather than learning genuine task completion.
Cost: $1.97 Per Task
On FrontierCode’s Main evaluation set, SWE-1.7 costs $1.97 per task on average. Cognition plots this on a cost-performance Pareto curve where SWE-1.7 occupies a position “that none of the pricier models occupy” — meaning you’d pay significantly more per task with GPT-5.5 or Opus 4.8 to get a small benchmark gain.
At 1,000 tokens/second via Cerebras, the model is also fast enough for interactive coding workflows rather than just batch background work.
The Caveat Builders Need to Know
SWE-1.7 is not an API model. You cannot call it directly.
It’s exclusively available through Devin — Cognition’s agentic coding product (web, desktop, and CLI). This is a deliberate product decision: Cognition is not selling model API access, they’re selling a coding agent service.
What that means for your stack:
- If you’re already using Devin for agentic coding tasks, you get SWE-1.7 automatically — it’s the new default model
- If you want to build agents with this architecture yourself, you need to wait for the techniques (RL-on-RL, self-compaction, entropy preservation) to appear in open-weights or API-accessible models
- Kimi K2.7 Code itself is open weights, so the base model is available — but SWE-1.7’s trained weights are not
What This Means for Builders Not on Devin
Even if you’re not a Devin user, this release matters for a few reasons:
The RL-on-RL insight will spread. If Cognition’s entropy preservation technique works, other labs will replicate it. Expect to see “extended RL post-training” appear as a differentiator in future model releases from Mistral, DeepSeek, and others working with open base models.
Self-compaction is an architecture builders should watch. The idea of a model that summarizes its own working state to extend effective task horizon is directly applicable to agentic system design — not just as a training technique but as an inference-time strategy. If you’re building long-horizon agents today, you’re doing a manual version of this with prompt engineering; future models may do it natively.
Benchmark positioning matters. SWE-1.7 at 81.5% Terminal-Bench makes the case that frontier-competitive agentic coding performance is no longer exclusive to the top-tier API models. If Cognition eventually opens an API (or if the techniques appear in open models), the competitive dynamics of the model routing market change.
Bottom Line for Builders
SWE-1.7 is a legitimate technical achievement. The RL-on-RL result challenges conventional assumptions about post-training ceilings, the self-compaction technique is novel, and the Terminal-Bench position (81.5% — above Sonnet 5) is credible at a claimed $1.97/task cost.
The access constraint is real: you get SWE-1.7 by using Devin, not by calling an API. If Devin fits your workflow, this upgrade is meaningful. If you’re building directly on model APIs, track the techniques rather than the model.
ChatForest covers AI model releases and builder implications as they happen. The benchmark data in this article is sourced from Cognition’s official SWE-1.7 announcement and cross-referenced against BenchLM.ai.