28B Model Upsets Big Tech: Pokee-Isaac Takes #1 in BFCL v4 Agentic Capabilities
Kyrannio · x · 2026-08-05
A user highlighted that the newly released Pokee-Isaac 28B scored 70.94 on the BFCL v4 agentic benchmark, narrowly beating GPT-5.6-luna (70.61) to take the top spot.
The model is noted as the only one to genuinely support a 10 million token context (where competitors score 0.0), while also demonstrating exceptional jailbreak resistance with a 35.6% ASR on the DTAP security test.
The author criticized major labs for charging $200/month for closed-weight models, arguing that a highly capable 28B model priced at $0.15/M input tokens proves that the industry's perceived moat was never really there.
Related event: Pokee-Isaac 28B Launches: 10M Context on Single GPU, Tops BFCL v4(5 posts)→
More from Models
- OpenAI's gpt-5.6-luna So Cost-Effective It Constantly Overloads Servers — _lewtun · 2026-08-05
- Opus 5 Language Degradation: Anthropic's Model Accused of Being a 'Jargon Douche' — TheTuringPost · 2026-08-05
- NVIDIA Releases Alpamayo 2 Super: A 34B Parameter VLA Model for Autonomous Driving — drmapavone · 2026-08-05
- AI Needs a Nap? Developer Exposes Hilarious GPT Thinking Block Hallucination — cantrell · 2026-08-05
- engyai Launches Cheapest Kimi K3 API on OpenRouter, Cutting Costs by 50% — const_reborn · 2026-08-05
- LiquidAI's LFM2.5-2.6B Hits 82 tok/s Decode on Mac with 128K Context — helloiamleonie · 2026-08-05