DeepSeek's MIT-licensed model just undercut Claude Code and Codex on price — by a lot

DeepSeek's retrained V4-Flash-0731 model beats its own flagship on every published agentic coding benchmark while costing a fraction as much per token, adding a credible open-weight option to the Claude Code vs Codex vs Cursor conversation founders are already having.

DeepSeek’s V4-Flash-0731 checkpoint went live on Hugging Face on 31 July 2026, MIT-licensed, and immediately outscored the company’s own larger V4-Pro-preview flagship on all nine agentic benchmarks it publishes — Terminal Bench 2.1 went from 72.1 to 82.7, and DeepSWE jumped more than sevenfold, from 7.3 to 54.4, from post-training alone. The API prices at $0.14 per million input tokens and $0.28 per million output tokens — roughly 3x cheaper than DeepSeek’s own Pro tier, and a fraction of what Claude Code or OpenAI Codex charge per token. It also ships with native OpenAI Responses API support, meaning it slots straight into Codex CLI without a custom integration layer.

Search interest around “AI coding agent pricing” and “cheapest AI coding tool” has been climbing steadily through 2026, and this release is exactly the kind of story that drives it: a credible, MIT-licensed, benchmark-leading model that costs a fraction of the market leaders. One important caveat — every one of those nine benchmark numbers is vendor-stated, run on DeepSeek’s own harness, with no independent reproduction confirmed as of this writing. Treat the headline figures the way you’d treat any vendor’s own marketing claims: directionally useful, not proof.

So what

Cheap, capable open-weight coding models widen the field beyond “which of the big three do we standardise on” — and that’s good news for anyone budgeting an AI-assisted build, because it puts real competitive pressure on Anthropic and OpenAI’s pricing. But price-per-token was never the hard part of shipping production software with AI agents; integration, code review discipline, and knowing which tasks actually benefit from agentic coding are. If you’re weighing which AI coding stack to commission a build on — and whether the cheapest option is actually the right one for your team — that’s exactly the kind of decision we help clients make. See our AI-assisted development approach or get in touch to talk through your setup.

← All research

More research.

GitHub secret scanning cuts false alarms for AWS, Google and Alibaba cloud keys and adds a SpaceX detector

GitHub's 5 November 2026 secret scanning update tightens detectors for AWS, Google Cloud and Alibaba keys, relevant to teams building software with AI coding tools that can paste credentials into code.

Read ›

GitHub's Copilot code review can now be managed by API, with review effort levels and AI credit costs shown

GitHub's 6 November 2026 update lets organisations manage Copilot code review by API and see the review effort and AI credits each AI code review used, a sign AI-assisted development is becoming auditable.

Read ›

GitHub's Copilot impact dashboard now counts chat and VS Code agent use, and drops the 'passive users' label

On 6 November 2026 GitHub added Copilot Chat and the VS Code Agents window to its Copilot impact dashboard, bringing it to nine tracked experiences, and renamed 'Passive users' to 'Chat-first users'.

Read ›

Claude Code 2.1.319: a setting that denies permission prompts so AI agent runs don't stall

Claude Code 2.1.319 (6 November 2026) adds a permissions.denyInsteadOfPrompting setting and moves long MCP tool calls to the background, a sign AI-assisted development is being built for unattended runs.

Read ›

GitHub now reports how often developers act on Copilot code review comments

On 5 November 2026 GitHub added code review metrics, including an addressed rate, to the Copilot usage metrics API. For AI software development buyers, it is the first first-party number for how much AI review feedback is acted on.

Read ›

Dependabot can now keep AI agent skills and prompts up to date

GitHub's Dependabot gained Agent Package Manager (APM) support on 3 November 2026, treating AI agent skills, prompts and instructions as dependencies that need updating like any other software.

Read ›

Start a project

Building with AI? Let's talk.

Tell us what you are building. We'll give you an honest view of the right approach.

Start a projectSee our work