CodeRabbit raises $143M as AI code review hits $1.5B valuation

CodeRabbit raises $143 million at $1.5 billion valuation for AI code review

CodeRabbit announced Wednesday that it raised $143 million in a round co-led by Atomico and Smash Capital, with BMW i Ventures and Datadog joining as new investors. Chief executive Harjot Gill gave the post-money valuation to Axios Pro as $1.5 billion. The company, founded in San Francisco in 2023 by Gill and Gur Singh, runs more than two million reviews a week for over 17,000 customers including Nvidia, BMW, Adyen, Indeed, JFrog and Trivago. Revenue grew more than five times year over year, though the base figure was not disclosed.

The company is introducing a product category called Agentic Change Management, arguing that issue tracking is dead because AI now produces code before anyone decides whether it is worth having. Three new tools follow: Triage scores incoming pull requests by value and risk, Change Stack groups large agent-written diffs into human-readable layers, and CodeRabbit Security scans code that has already shipped. The company plans to spend $10 million over 12 months keeping the tool free for open-source maintainers, covering 150,000 projects. It has opened an office in London and employs 50 people across the UK and EU, with plans to double that headcount and expand into Japan and Singapore.

CodeRabbit raised $143m to read the code AI wrote →

Google launches Gemini 3.7 Flash with big coding gains at half the price

Google released Gemini 3.7 Flash just three weeks after Gemini 3.6 Flash, calling it its most capable workhorse model yet for coding and AI agents. Google attributed the rapid intelligence gains to algorithmic improvements. On FrontierCode, the model scores 43.6 percent, up from 34.4 percent, and on DeepSWE it scores 65.3 percent, up from 49.0 percent. According to Google’s measurements, it outperforms Claude Sonnet 5 and GPT-5.6 Terra, with gains also in web development, document comprehension, and business process automation.

The model is available through the API, AI Studio, and Antigravity. Launch pricing is $0.75 per million input tokens and $3.75 per million output tokens, 50 percent cheaper than 3.6 Flash at launch. Both models now share the same price point, and the pricing holds through the end of the year.

Gemini 3.7 Flash lands with coding gains and undercuts its three-week-old predecessor’s price by 50% →

Deepseek ships V4 Pro, open-sources agent software, and raises API prices

Deepseek moved its flagship model out of testing, released its agent software as open source, and announced higher API prices effective August 16. The deepseek-v4-pro endpoint now delivers build V4-Pro-0813, with the model name, parameter count, and one-million-token context window unchanged. Terminal Bench 2.1 scores jumped from 72.1 to 87.9, and DeepSWE scores went from 12.8 to 62.7. On several agent benchmarks, the model beat Claude Opus 4.8, though Artificial Analysis places V4-Pro at 53 on its Intelligence Index, behind Muse Spark at 57, Qwen 3.8 Max at 58, Kimi K3 at 60, and Claude Opus 5 at the top with 63.

The company also shipped Deepseek Harness v0.1 as a Developer Preview under the MIT license, an open-source agent software alternative to OpenAI’s Codex and Claude built on the Cordis plugin system. The new pricing introduces peak and off-peak rates, with off-peak usage costing half as much. During off-peak hours, V4-Pro input goes from $0.435 to $0.66 per million tokens and output from $0.87 to $1.98; peak hours double those rates. Cache hits see the steepest increase, from $0.003625 to $0.022 off-peak and $0.044 at peak, shrinking the cache discount significantly. The price hike comes as the company is raising new capital and preparing for an initial public offering.

Deepseek ships improved V4 Pro, open-sources its agent software, and raises API prices →

OpenAI report finds no correlation between AI usage and revenue per employee

OpenAI published a 69-page report on August 11 on enterprise adoption of ChatGPT, describing exponential AI usage growth across seniority levels and job functions. However, a table on page 35 shows no statistically significant correlation between revenue per employee and AI usage, measured in messages sent and tokens used. The report states that revenue per employee is not meaningfully associated with output tokens per employee or messages per active user once other controls are included. Companies with higher revenue per employee tend to be early ChatGPT adopters, but the study does not establish that more AI use causes higher revenue.

The report also shows executives use AI the least intensely, with the lowest weekly messages per user. Enterprise usage flatlined from approximately October 2025 to December 2025 before becoming exponential in January 2026. Two of the report’s five authors are academics paid by OpenAI, a fact disclosed in a footnote. OpenAI CFO Sarah Friar wrote on LinkedIn that competitive advantage comes from the people closest to the work. The report comes as CEO Sam Altman reorganizes the company around enterprise sales and hires Dali Rajic as chief revenue officer ahead of a planned IPO.

Buried in OpenAI’s latest research: No correlation between AI use and revenue per employee →

Spotify to label AI creators posing as human artists

Spotify will begin rolling out AI Persona badges mid-September, appearing on artist profile pages, in related search results, and in playlist track listings. The badge addresses whether a profile represents an actual human and concerns the artist’s public identity, not how the music was made. Starting this week, creators can self-disclose if they use AI for their public persona, and Spotify will independently review profiles to determine whether names, identities, or images appear to be products of generative AI.

Users identified as AI Personas will be notified and can appeal the label. AI Personas and their music will be excluded from all Spotify recommendation features, including algorithmic and editorial playlists, and will appear on a listener’s feed only if followed or intentionally searched for. The move follows last summer’s The Velvet Sundown incident, a generative AI project posing as a human band that accumulated millions of streams in weeks. In April, Spotify introduced verification badges confirming an artist’s identity and authenticity.

Spotify will now label AI creators posing as human artists →