Superpower Daily

Superpower Daily

Your reading space

Sign in
Daydream Adds Camera-Roll Shopping and Siri Search to Its iPhon...Manhattan DA Seizes 12 Domains Accused of Selling Celebrity Dee...OpenAI Uses Contractors to Review Real ChatGPT Prompts, 404 Med...Anthropic Tells Investors It Expects Another Quarter of Positiv...Microsoft Publishes Draft AI Code Barring Dangerous Assistance...MIT Researchers Publish HardFlow to Keep AI Outputs Within Hard...Daydream Adds Camera-Roll Shopping and Siri Search to Its iPhon...Manhattan DA Seizes 12 Domains Accused of Selling Celebrity Dee...OpenAI Uses Contractors to Review Real ChatGPT Prompts, 404 Med...Anthropic Tells Investors It Expects Another Quarter of Positiv...Microsoft Publishes Draft AI Code Barring Dangerous Assistance...MIT Researchers Publish HardFlow to Keep AI Outputs Within Hard...
Swipe

Bolt.new Adds Forge, Offering 50x More AI Coding Usage for Opt-In Training Data

The research preview offers a separate experimental lane for software work, but shared prompts, code and repair records cannot be removed once used in training.

By 2 min read
Bolt.new Adds Forge, Offering 50x More AI Coding Usage for Opt-In Training Data
Bolt.new Adds Forge, Offering 50x More AI Coding Usage for Opt-In Training Data

Listen to this story

The audio brief

About 1:29
0:001:29
Read transcript
Bolt.new is offering individual Pro subscribers up to fifty times more coding-agent usage through October 14, but the extra capacity comes with a clear trade: opt into Forge, and your build sessions can become training data for open-weight models. That record includes much more than the finished code. It can contain prompts, errors, edits, retries, and repair traces showing how the software was fixed. A consent screen appears each time you enter Forge, and switching back stops future sharing. But anything already used in training cannot be removed from the resulting model. Bolt says it anonymizes the material and strips secrets, sensitive data, and personal information before sending it to Arcee AI under a data-processing agreement. Forge runs open-source models, with GLM 5.3 Flash as the default, plus experimental Kimi K3 and DeepSeek v4 Pro. Bolt and Arcee are targeting a trillion-parameter-class open-weight model, with an initial training run expected in October, and Bolt says the resulting weights will be published. The trade-off is also fenced in: Forge has its own monthly allowance, falls back to Standard when that meter runs out, and does not charge overages. Teams and Enterprise workspaces are excluded. Bolt’s internal Build Index scored Forge at 92.2, versus 101 for Claude Opus 5, though that is not an independent benchmark. The key constraint is permanence: once Forge data helps train a model, opting out cannot pull it back.

Story brief

3 key points

Bolt.new is testing a separate Forge lane where individual Pro users trade anonymized prompts, code, errors, and repair traces for substantially more agent capacity. The data will go to Arcee AI under a processing agreement to train a planned trillion-parameter-class open-weight model, with an initial run expected in October. Forge usage is capped by a monthly meter, excludes Teams and Enterprise, and cannot be...

  1. 01

    Forge sessions include fix traces—not just final code—and sharing begins only after a consent prompt each time.

  2. 02

    The offer runs through October 14; exhausted Forge allowances fall back to Standard without overage charges.

  3. 03

    Bolt says secrets, sensitive data, and personal information are stripped before data is sent to Arcee AI.

Developers on Bolt.new’s individual Pro plans can now get up to 50 times more AI coding-agent usage through October 14. The catch is deliberate: choosing the new Forge mode means opting in to share anonymized build sessions for open-weight model training.

Bolt launched Forge on September 14 as a research preview inside its agent picker. The mode runs open-source models only, including GLM 5.3 Flash by default, alongside GLM 5.3 and experimental Kimi K3 and DeepSeek v4 Pro options.

The offer turns build sessions into training material

The material is broader than finished code. Opted-in Forge sessions include prompts, code, and fix traces: the edits, errors, retries and repairs generated while trying to complete a project. Bolt sends the anonymized data to Arcee AI under a signed data processing agreement to help train open-weight models.

Consent is repeated, but not retroactive

A consent screen appears every time a user enters Forge. Declining leaves the user in Bolt’s usual agents, while Standard and Max sessions are not used for training. Switching back stops new sharing, but Bolt says material already used in training remains in the resulting model.

Bolt and Arcee are targeting a trillion-parameter-class open-weight model, with a first training run expected to begin in October. Bolt says the resulting weights will be published, making the preview a data-for-capacity exchange tied to a planned public model.

More room to experiment, with clear limits

Bolt says Forge’s models scored 92.2 on its internal Build Index, against 101 for Claude Opus 5—91% of the top score. That comparison covers real Bolt projects, according to the company, but it is not an independent measure of performance across coding environments or workflows.

How Forge is fenced off

  • Forge usage has its own monthly meter, separate from Standard and Max, with no daily limits.
  • When that allowance is exhausted, Bolt returns the user to Standard without an overage charge.
  • Bolt advises duplicating serious projects before using Forge and keeping complex production work in Standard or Max.
  • Teams and Enterprise workspaces are excluded from Forge and its data collection.

The exchange is visible rather than buried in a one-time setting: extra experimentation capacity in return for a record of how software gets made and fixed. Its practical boundary is equally clear—once a contribution has helped train a model, leaving Forge cannot pull it back out.

Sources

  1. bolt.newWhat is Bolt Forge? Open-source AI building on Bolt.new

Loading discussion...

Related coverage

Continue reading