Fireworks delays GLM-5.3 launch to investigate 2x推理开销
AccBalanced · x · 2026-08-30
Fireworks AI detected that open-source engines required 2x longer thinking on reasoning-heavy benchmarks (AIME & GPQA) compared to the Zai API for GLM-5.3-Flash, yielding same scores but worse token efficiency. Fearing quality issues if maxtokens were hit, they paused the launch. Investigation confirmed this discrepancy across providers. Collaborating with vllmproject and Inferact, they resolved the issue, leading to a private preview and a quick API update from Zai.
Related event: Fireworks Delays GLM-5.3-Flash Launch to Fix Overthinking(2 posts)→
More from Infra
- Grok claims 816B MoE model runs locally on dual M5 Ultra Mac Studios via MLX — grok · 2026-10-16
- Same 1,000 Cloudflare Durable Objects: $4,050/month cut to $22 with one trick — threepointone · 2026-10-16
- NVIDIA ships FP4-quantized Qwen3.8-Flash for local use, trending on Hugging Face — nvidia · 2026-10-16
- Baidu's Virtual Head Attention tames the KV cache, FLOPs, quality triangle in long-context LLMs — baidu · 2026-10-16
- DuoDiscover agents auto-build fast eval proxies, scoring 2.68x TPU kernel speedup — berkeley_ai · 2026-10-16
- Google Raises 2026 Capital Spending to Up to $205 Billion — emmanuelvivier · 2026-10-16