Kimi K2.7 post-trained on 2,400 coding tasks lifts DeepSWE from 31% to 48.7%
echen · x · 2026-10-22
New post-training run: Kimi K2.7 trained on 2,400 coding tasks gained +17.7pp on DeepSWE (31%→48.7%), +14.3pp on Terminal Bench 4 (0%→14.3%), and +8.0pp on SWE-Bench Pro (60.1%→68.1%). Results are published on Surge AI's Post-Training Runs platform, which also shows Nemotron Super 3.5 gains from Surge data: +11.5pp AutomationBench, +14.6pp Chartography, broad instruction-following improvements, and +7.9pp DeepSWE, with analysis of transfer beyond the training distribution.
Related event: Kimi K2.7 Post-Training on 2,400 Coding Tasks Boosts Benchmarks(2 posts)→
More from Models
- Recursive AI ships Flint, a 40B-A2.5B MoE claimed to match Qwen3.6-27B, weights on Hugging Face — ChengleiSi · 2026-10-22
- DeepSeek ships V4.1-Flash-1017, now the #1 open model on RSI-Inference Bench — scaling01 · 2026-10-22
- OpenAI's Cybersecurity Guardrails Block a Dev From Auditing His Own Patched Server — Next_Technology6361 · 2026-10-22
- TPN-006 hits Hugging Face: 7.7GB compressed model beats its 17GB Mistral teacher — markjeffrey · 2026-10-22
- Kimi K3.1's latest checkpoint draws 49k-token SVG portrait zero-shot in 22 minutes — lyraxana · 2026-10-22
- MiniMax launches M3.1 Flash at 1/3 the cost and 3x the speed of M3, weights out Oct 23 — MiniMax_AI · 2026-10-22