Moonshot AI published Kimi K2.7-Code to Hugging Face on June 12 (MarkTechPost), a 1-trillion-parameter Mixture-of-Experts model with 32 billion active parameters across 384 experts per forward pass and a 256K-token context window (Hugging Face). Released under a Modified MIT license and designed for long-horizon agentic software engineering, the model reduces reasoning-token consumption by approximately 30% compared to Kimi K2.6 (Crypto Briefing). Moonshot reports a 21.8% gain on Kimi Code Bench v2 and a 31.5% improvement on MLS Bench Lite over the prior version, and scores 81.1% on MCPMark Verified against Claude Opus 4.8’s 76.4% on the same benchmark (MarkTechPost). API access is priced at $0.95 per million input tokens and $4.00 per million output tokens via the model ID kimi-k2.7-code (Crypto Briefing).
Anthropic made Claude Fable 5 publicly available on June 9, marking the first Mythos-class model released to general users, with reported scores of 95% on SWE-bench Verified and 80% on SWE-bench Pro (Enterprise DNA, TechCrunch). The model is available at no extra cost on Pro, Max, Team, and Enterprise plans through June 22, after which Fable 5 and Mythos 5 will be priced at $10 per million input tokens and $50 per million output (CNBC); in high-risk domains such as cybersecurity, biology, and chemistry, Fable 5 blocks responses and falls back to Claude Opus 4.8 (The Hacker News). On June 1, GitHub transitioned all Copilot plans to usage-based billing measured in AI Credits priced at $0.01 each, converting each subscription tier 1:1 to included monthly credits - Copilot Business at $19/user/month includes $19 in credits, Copilot Enterprise at $39/user/month includes $39 - with token consumption metered across input, output, and cached tokens (GitHub Blog, GitHub Changelog).