Anthropic published two official patterns to run its most powerful model, Claude Fable 5, at half the price by using cheaper models for routine work and reserving Fable for judgment calls. The advisor…
MiniMax, a Chinese AI lab, open-sourced M2.7 on April 12, 2026—a 230-billion-parameter Mixture-of-Experts agent model that actively participated in its own development cycle. During training, M2.7 had…
Meituan released LongCat-2.0, a 1.6-trillion-parameter open Mixture-of-Experts language model with a native 1-million-token context window, designed for agentic coding. The model activates about 48 bi…
Anthropic released Claude Sonnet 5 on June 30, 2026, offering near-Opus performance at significantly lower cost, scoring 63.2% on SWE-bench Pro and 1,618 on GDPval-AA v2. Priced at $2 per million inpu…
Anthropic launched Claude Sonnet 5 on June 30 with a two-month introductory pricing of $2 per million input tokens and $10 per million output tokens, set to increase 50% on September 1 to $3/$15. The …
A new Cursor study found that 63% of successful Opus 4.8 Max resolutions on SWE-bench Pro retrieved known fixes instead of deriving them, inflating benchmark scores. Sealing git history and internet a…
Nous Research's open-source Hermes Agent, using Mixture of Agents presets, outperformed Anthropic's Claude Opus 4.8 and OpenAI's GPT-5.5 on SWE-bench Pro benchmarks. The framework strings multiple lan…
A new study finds that 63% of successful Opus 4.8 Max resolutions on SWE-bench Pro retrieved the fix rather than deriving it, with scores dropping sharply when git history and internet access were res…
Z.AI's open-weight GLM-5.2 model outperforms GPT-5.5 on the SWE-bench Pro coding benchmark, scoring 62.1 versus 58.6, while costing $1.40 per million input tokens compared to GPT-5.5's $8.00. Released…
Anthropic's Claude Opus 4.5 and Zhipu AI's GLM-5.2 are frontier reasoning models competing on cost, context length, and coding benchmarks. GLM-5.2 offers a 1M-token context window and leads by 20.3 po…
OpenAI abandoned SWE-bench Verified on February 23, 2026, after finding 59.4% of its hardest failed tests were broken and training data contamination inflated scores. Its replacement, SWE-bench Pro fr…
Z.ai's MIT-licensed GLM-5.2 model challenges OpenAI's GPT-5.5 with a 28% hallucination rate on the AA-Omniscience benchmark, three times lower than GPT-5.5's 86%, while costing 1/7th the inference pri…
Zhipu AI released GLM-5.2, an open-source large language model with a 1M-token context designed for long-horizon coding tasks. The model introduces IndexShare to reduce computational costs and achieve…
Xiaomi released MiMo Code V0.1, an open-source AI coding agent, on June 11 under an MIT license, achieving an 86.7% score on Terminal-Bench 2.0 compared to Claude Code's 65.4% on tasks exceeding 200 s…
Anthropic released Claude Opus 4.8 on Tuesday, pricing the new model at $5 per million input tokens and $25 per million output tokens — the same as its predecessor. The company claims the model is its…
Major developments in AI development tools as of April and May 2026, highlighting Cursor 3.0's shift to a multi-agent workspace, Anthropic's Claude Code achieving 87.6% on SWE-bench Verified, and Wind…