Qwen3.8-Max rapidly gained popularity on B.AI, driving daily token throughput to 14.35 billion.

// 1h agoNEWS

Qwen3.8-Max rapidly gained popularity on B.AI, driving daily token throughput to 14.35 billion.

B.AI released platform insights revealing that Qwen3.8-Max quickly became one of its most popular models. The surging demand pushed daily token throughput to 14.35 billion tokens with API usage reaching 99.7% capacity.

// ANALYSIS

High demand for Qwen models on multi-model platforms highlights the growing developer preference for capable open-weights models and diverse API access.

–Daily token throughput surged to 14.35 billion.
–API infrastructure operated at near-maximum capacity (99.7%).
–Qwen3.8-Max rapidly gained strong developer adoption on the platform.

// TAGS

b.aiqwen3.8-maxllmaitoken-throughputapi

DISCOVERED

1h ago

2026-08-09

PUBLISHED

2h ago

2026-08-09

RELEVANCE

7/ 10

AUTHOR

Dalat_Ventures

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

OPEN SOURCE2h ago

Codex Self-Improving Loop Uses DeepSeek to Refine Prompts

codex-self-improving-loop is an open-source workflow framework designed to run iterative self-improving agent cycles within Codex using DeepSeek. The system uses a multi-agent architecture where DeepSeek executes tasks from prompt.md, an evaluator role (Luna) grades each run, and a prompt-rewriter role (Sol) updates the prompt file based on feedback until completion.

UPDATE2h ago

Synara enables importing AI conversation threads across platforms

Synara has added a new capability that lets users import conversation threads from other AI providers directly into its application interface. This feature allows developers to preserve task history and maintain context continuity when transitioning between different AI assistants without having to restart long-running agent sessions from scratch.

UPDATE2h ago

Octra introduces continuous reasoning and validator ML verification

Octra shared key technical updates regarding its private AI and state inference architecture in its latest weekly recap. The network now enables continuous reasoning without needing to clear circle memory, allowing AI models and applications to maintain state across execution steps. In addition, validators can now verify useful compute workloads including machine learning inference and text generation, expanding the scope of verifiable private compute on-chain.