DeepSeek released V4.1-Flash-1017, which outperforms GPT 5.6 Sol on Terminal-Bench while costing about 17x less. The gain comes from RL training across multiple…
DeepSeek V4.1 Flash 1017 outperforms Claude Opus 5 on DeepSWE v1.1 by about 3 points while costing roughly 60x less, and is also 2-2.5x cheaper than the prior D…
DeepSeek V4.1 Flash 1017 scores 77.0 on DeepSWE versus Opus 5's 74.0, at roughly 20 cents per task, and is live in Command Code.
Ahmad Awais announced that DeepSeek V4.1 Flash 1017 is now live on Command Code under the same existing alias.
Qoder announced that DeepSeek V4.1 Flash has been updated to version V4.1-Flash-1017, and the entire Qoder suite now supports the new version.
DeepSeek V4.1 Flash 1017 is now live on OpenRouter, with DeepSeek as the first provider and more providers coming. The update is more token efficient and shows…
DeepSeek released DeepSeek-V4.1-Flash-1017 with stronger benchmark performance, faster task completion and shorter outputs than the previous version, now live o…
DeepSeek launched DeepSeek-V4.1-Flash-1017 with better benchmarks, faster tasks and shorter outputs at lower cost, and opened its Search & Fetch APIs at $0.75 p…
Zed published a video in which its Delta agent configures itself to use Ollama as a custom OpenAI-compatible provider entirely within the thread, without the us…
Replicas announced that Ollama is now an inference provider on its platform, letting users run open models inside tools such as Claude Code, Codex, OpenCode and…
