FrontierCode 2.0: Anthropic and OpenAI lead on intelligence-per-dollar, not just price
brandon_galang · x · 2026-10-28
brandongalang argues FrontierCode 2.0 is the clearest counter-signal to FUD about Chinese OSS models threatening Anthropic and OpenAI. The leaderboard shows Opus 5.5 max on top at 45.5% (49.7% pass rate, $12.82/rollout), GPT-6 Astra max at 42.8% ($6.68), Sonnet 5.5 max at 41.0%, while DeepSeek V4.1 Flash (19.3%, $0.38) and GLM 5.3 (17.8%, $7.52) sit at #16-17.
His core claim: price-per-token is a meme — the real game is intelligence and productivity per dollar, where Anthropic and OpenAI hold a commanding lead.
More from Models
- Gym-Anything-RT: best frontier model solves only 9% of real-time tasks vs 45% for humans — wellecks · 2026-10-28
- NVIDIA's Nemotron 3.5 Transcribe ties for #1 open-weight STT at 2.8% WER with just 4B active params — ArtificialAnlys · 2026-10-28
- An Interesting Real-World Use Case of Genie 4 — MicahBerkley · 2026-10-28
- Developer says Opus/Codex slide decks are so recognizable he's grown to hate them — felpix_ · 2026-10-28
- Dev to Anthropic: Claude should text like a friend, not email like a assistant — weswinder · 2026-10-28
- NVIDIA Releases Nemotron 3.5 Transcribe and Granary v2 for Multilingual ASR — Hugging Face Blog · 2026-10-28