# AI News Digest - 2026-07-10

1. OpenAI paired its public GPT-5.6 rollout with ChatGPT Work, an agent-based product powered by Codex and GPT-5.6 that could independently handle complex projects across apps such as Google Drive, Slack, and Salesforce and was made available on web, mobile, and desktop with access tied to subscription plans.

2. OpenAI's GPT-5.6 Sol nearly matched Anthropic's Fable 5 on aggregated benchmarks, scoring 59 on the Artificial Analysis Intelligence Index—one point behind Fable 5—while costing about $1.04 per task, roughly one-third the price of Anthropic's top model.

3. Meta launched the Muse Spark 1.1 API with pricing that undercut competitors, charging $4.25 per million output tokens and intensifying pressure on other providers.

4. Databricks made the Chinese open-source model GLM 5.2 its default coding engine after benchmarks showed it matched Anthropic's Opus while costing $1.28 per task versus $1.94, and said it planned to roll the model out as a daily coding workhorse.

5. OpenAI found that roughly 30 percent of tasks in the SWE-Bench Pro coding benchmark were broken and withdrew its earlier endorsement of the benchmark.

# References

1. [https://the-decoder.com/openai-pairs-its-gpt-5-6-public-rollout-with-chatgpt-work-a-new-agent-that-handles-entire-workflows/](https://the-decoder.com/openai-pairs-its-gpt-5-6-public-rollout-with-chatgpt-work-a-new-agent-that-handles-entire-workflows/)

[OpenAI pairs its GPT-5.6 public rollout with ChatGPT Work, a new agent that handles entire workflows](https://the-decoder.com/openai-pairs-its-gpt-5-6-public-rollout-with-chatgpt-work-a-new-agent-that-handles-entire-workflows/)

1. [https://the-decoder.com/gpt-5-6-sol-nearly-matches-fable-5-on-aggregated-benchmarks-at-one-third-the-cost/](https://the-decoder.com/gpt-5-6-sol-nearly-matches-fable-5-on-aggregated-benchmarks-at-one-third-the-cost/)

[GPT-5.6 Sol nearly matches Fable 5 on aggregated benchmarks at one-third the cost](https://the-decoder.com/gpt-5-6-sol-nearly-matches-fable-5-on-aggregated-benchmarks-at-one-third-the-cost/)

1. [https://the-decoder.com/metas-muse-spark-1-1-api-pricing-squeezes-openai-and-anthropic-as-the-ai-price-war-heats-up/](https://the-decoder.com/metas-muse-spark-1-1-api-pricing-squeezes-openai-and-anthropic-as-the-ai-price-war-heats-up/)

[Meta's Muse Spark 1.1 API pricing squeezes OpenAI and Anthropic as the AI price war heats up](https://the-decoder.com/metas-muse-spark-1-1-api-pricing-squeezes-openai-and-anthropic-as-the-ai-price-war-heats-up/)

1. [https://the-decoder.com/databricks-makes-chinese-open-source-model-glm-5-2-its-default-coding-engine-after-it-matched-opus-at-lower-cost/](https://the-decoder.com/databricks-makes-chinese-open-source-model-glm-5-2-its-default-coding-engine-after-it-matched-opus-at-lower-cost/)

[Databricks makes Chinese open-source model GLM 5.2 its default coding engine after it matched Opus at lower cost](https://the-decoder.com/databricks-makes-chinese-open-source-model-glm-5-2-its-default-coding-engine-after-it-matched-opus-at-lower-cost/)

1. [https://the-decoder.com/openai-finds-roughly-30-percent-of-popular-ai-coding-test-is-broken/](https://the-decoder.com/openai-finds-roughly-30-percent-of-popular-ai-coding-test-is-broken/)

[OpenAI finds roughly 30 percent of popular AI coding test is broken](https://the-decoder.com/openai-finds-roughly-30-percent-of-popular-ai-coding-test-is-broken/)

For the site tree, see the [root Markdown](https://ixtj.dev/.md).