Gemini 4 Carbon beats Claude Opus 5.5: 23% lower cost per task, #1 on Intelligence Index
DeepLearningAI · x · 2026-11-10
DeepLearning.AI's newsletter The Batch compared Gemini 4 Carbon vs. Claude Opus 5.5:
- Same price per token (excluding Carbon's intro discount)
- Carbon uses 29% fewer output tokens per task
- 23% lower overall cost per task
- Carbon ranks #1 on the independent AI Intelligence Index (59)
The attached ExploitGym chart (causal pass rate, time-capped, higher is better) shows Gemini 4 Carbon (10h) at 54.8% and its 6h variant at 53.5%, ahead of GPT-6 Astra (42.4%), Gemini 4 Argon (42.0%), GPT-6.1 Sol (35.1%), Claude Opus 5.5 (34.5%), Claude Mythos 5.1 (30.4%), with MiMo-V2.6-Pro at 17.8% and GLM-5.3 at 15.0%. Carbon's efficiency edge comes from spending fewer tokens, not a cheaper unit price.
More from Models
- GPT-6 Astra edits Qwen3.5-0.8B weights directly: 0.016% of params, MATH-500 from 7.9% to 48.1% — sebkrier · 2026-11-10
- Fable 5.2 launch offers: up to 3 usage resets for long-time subscribers, one-week trial for Pro — ClaudeDevs · 2026-11-10
- Fable 5.2 effort levels: API defaults to high, xhigh reserved for long-running agentic work — ClaudeDevs · 2026-11-10
- Anthropic extends preserved thinking to all API accounts on Fable 5.2 to block distillation — ClaudeDevs · 2026-11-10
- Matthew Berman: Fable 5.2 has 'completely solved' motion graphics — one-shot Fall Guys clone included — MatthewBerman · 2026-11-10
- Claude Fable 5.2 tested: one prompt turns 5GB of federal data into a US water-safety map — alliekmiller · 2026-11-10