AI Coding Live News

bolt.new introduced Bolt Forge, an open-weight coding agent now live in the bolt.new model picker. Free until Oct 14 with up to ~50x usage and zero usage charges; lineup includes GLM 5.3 / GLM 5.3 Flash, Kimi K3, and DeepSeek v4 Pro. Opt-in builds help train a trillion-parameter-class model with Arcee; also announced $9/mo Bolt Lite (Forge included; join by Oct 14 to lock $9). Partners include Arcee, Microsoft, Vercel, Fireworks, and DigitalOcean.

Fireworks said partner LangChain generates billions of agent-trace tokens a day, making frontier closed judges too costly. Fine-tuning a Qwen base on Fireworks matched GPT-5.5 / Opus-level judging at up to about 100x lower cost, and Fireworks is pushing its Training API for similar workflows.

Fireworks said DeepSeek V4 Flash now includes native vision on serverless and priority tiers. It reports better benchmarks than the text-only 0731 checkpoint at the same $0.22 input / $0.66 output per 1M tokens. The model page lists a 305B MoE with about 1040K context, function calling, image input, and a speculative-decoding module.

Fireworks announced general availability of its Training API and Fireworks Lab, aiming to make model specialization accessible beyond labs that already own training clusters. The API lets teams connect a custom Python training loop to Fireworks-managed distributed training and rollout infrastructure, keeping control of loss, reward, data, and environment. Compute is split into Serverless mode for fast experiments and Dedicated mode when scale, control, and GPU economics matter. The stack is built for reinforcement learning as a single train-sample-serve-retrain loop, with three hard requirements: rollout throughput, weight sync so fresh weights land on the rollout deployment every step, and correctness so trainer and inference engine agree on the same model. Fireworks Lab additionally embeds forward-deployed researchers and engineers to co-design or build a model. Named customers include Cursor, Harvey, Figma, Vercel, Cognition, Factory, and Heidi.

Fireworks spotlighted Factory's Droid Shield as a Training API customer story. Factory's droids write and commit code faster than human reviewers can keep up, so secret detection has to catch real leaks without flooding the team with false alarms. Using the Fireworks Training API, Factory fine-tuned an open Qwen model that caught almost 20 percent more real secrets than GPT-5.5, at lower cost and latency. The result is a specialized scanner sitting in the commit path rather than a generic frontier chat model asked to grep for keys. For coding-agent vendors, it is a concrete example that a domain-fine-tuned open-weight model can beat a larger closed model on a verifiable production task, then ship at a cost that matches high-frequency agent traffic. It also shows why training and inference on one platform matters: the specialized weights have to stay cheap enough to run on every commit.

GitHub said Gemini 3.7 Flash, Microsoft MAI-Code-1.1-Flash, and the Fireworks-hosted open-weight Kimi K3 model are generally available in the GitHub Copilot app, Copilot CLI, VS Code, and related surfaces. Developers can switch among Google's low-cost coding model, Microsoft's in-house code model, and an open-weight option without leaving the Copilot workflow. The same Copilot weekly thread also promoted a Customize tab that gathers MCP servers, plugins, skills, and canvases, plus Slack and Microsoft Teams integrations that let teams mention GitHub to plan changes, investigate problems, and hand off coding tasks from chat. For organizations standardizing on GitHub as the agent control plane, the release is less about a single SOTA model and more about routing: pick Gemini 3.7 Flash for cheap iteration, MAI-Code for Microsoft-stack tasks, or Kimi K3 when open weights and Fireworks hosting matter.

GitHub said Gemini 3.7 Flash, Microsoft MAI-Code-1.1-Flash, and the Fireworks-hosted open-weight Kimi K3 model are generally available in the GitHub Copilot app, Copilot CLI, VS Code, and related surfaces. Developers can switch among Google's low-cost coding model, Microsoft's in-house code model, and an open-weight option without leaving the Copilot workflow.

Fireworks AI launched day-zero US-hosted serverless inference for GLM-5.3, highlighting a 50% improvement on Code Bench and state-of-the-art results in open-weight cyber defense and long-horizon agents.

Fireworks AI detailed Harvey's proprietary model Tenet, post-trained from Kimi K3 using async RL on Fireworks Training API, doubling LAB Contract all-pass rates from 10.8% to 19.7% at matched cost.