GPT-5.6 Still Generates Too Many Reasoning Tokens
bdsqlsz · x · 2026-07-10
The author's comparison tests suggest that Luna is roughly equivalent to GPT-5.5 mini, Terra to GPT-5.5 medium, and Sol performs the best—with even lower-tier Sol outperforming the highest-tier Terra.
The quote adds that while GPT-5.6 sol is "smarter," it still suffers from excessive reasoning token generation. Unmodified, it consumes 518n-2 reasoning tokens even when the answer is correct. After the author's modifications, it used fewer tokens and achieved a 5/5 accuracy rate.
Related event: GPT-5.6 Release Sparks Discussion on Performance and Cost(10 posts)→
More from Models
- Daily AI brief: GPT-Live-1 in API, OpenAI pauses $200 Pro signups amid Astra demand — koltregaskes · 2026-09-11
- Same Echo Maze prompt, three frontier models: all passed visually but shipped the same hidden bug — eyishazyer · 2026-09-11
- Benchmark scores drop from 89% to 19% on new evals — how benchmaxxing breaks leaderboard trust — airesearch12 · 2026-09-11
- ChatGPT tells user their question is too hard and to 'accept dumber answers' — phido3000 · 2026-09-11
- Developer Building a Unified Leaderboard of All Model Benchmark Scores — airesearch12 · 2026-09-11
- Rumor claims Kimi faked performance by serving Claude; DeepSeek new model surprises in evals — realsohamparekh · 2026-09-11