Gemini 3.7 Flash lands with coding gains at half the price

Google's Flash line is now on a three-week release cycle, and its newest iteration arrives with meaningfully better coding scores and a price tag that undercuts its own predecessor by 50 percent.

Google shipped Gemini 3.7 Flash on Thursday, three weeks after Gemini 3.6 Flash, calling it its "most intelligent workhorse model yet for coding and agents" — and pricing it at half of what the previous Flash charged at launch. The model scores 43.6 percent on Google's FrontierCode benchmark, up from 34.4 percent for 3.6 Flash, and 65.3 percent on the DeepSWE software-engineering benchmark versus 49.0 percent — results Google says put it ahead of both Claude Sonnet 5 and GPT-5.6 Terra on its own measurements. The gains extend beyond code: Google reports improvements in web development, document comprehension, and business process automation, and credits "awesome algorithmic improvements" rather than a bigger model for the jump in intelligence. The company also says the new Flash ships with stronger safeguards against misuse in the chemical, biological, radiological, and nuclear domains and in cyber offense.

The headline number, though, is the price. Gemini 3.7 Flash launches at $0.75 per million input tokens and $3.75 per million output tokens — exactly half of what 3.6 Flash cost when it debuted three weeks ago — and Google says the rate holds through the end of the year, with both Flash models now sharing the same price point. For agent builders, that math matters: agents burn tokens at a furious rate, and halving the marginal cost of a task is the difference between an agentic feature that is economically viable at scale and one that is not. The cuts also land on the same day DeepSeek pushed its own API prices up — we covered DeepSeek hikes API prices up to 4.7x with new peak-hour billing earlier today — a reminder that pricing pressure in this market runs in both directions.

The cadence is the other story. Three weeks between Flash generations points to a production line Google has clearly industrialized, and it is doing so while its flagship Gemini 3.5 Pro still hasn't shipped — the workhorse, not the flagship, is where Google is choosing to fight right now. Flash is the model most agentic apps will actually run on, and the company is betting that volume plus algorithmic efficiency beats the premium-tier arms race. The model is available through the API, AI Studio, and Antigravity, and consumer access arrives via Spark, Google's agent, for AI Ultra and Pro subscribers.

What to watch: whether OpenAI and Anthropic answer with price cuts of their own, and whether Google can hold a three-week Flash cadence through the rest of the year.

At $0.75 per million input tokens, does Flash's price-to-performance edge change which model you build on? Tell us in the comments.

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Kuaishou's Kling AI has spent two years becoming the video model people actually finish projects with. Its new flagship, Kling 4.0, went fully live on October 15 — and this round is less about prettier frames than about the two things that make a generated shot usable in a real edit: length and control. Kling 4.0 generates up to 30 seconds in a single native pass, outputs 10-bit HDR at 4K, 1080p and 21:9, and takes up to 15 reference assets in one task. That budget splits across ten images, fi

Brain organoids classified Braille at 83% — in a trio, not alone

Two stories today: human brain tissue on a chip decoded a tactile signal, and the data-centre company behind some of the largest AI contracts in Europe turns out to be six months behind on its flagship site. Human forebrain organoids grown on a microelectrode array classified Braille letters with 61 percent accuracy on their own and 83 percent when the responses of three organoids were combined, in a peer-reviewed study from the University of Bristol. The team took event-based tactile readings

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute

Microsoft finished pre-training its biggest in-house model yet — then used it to make a point about the company it bankrolls. Microsoft AI completed pre-training of MAI-Base-2, a mixture-of-experts foundation model with 80 billion active and 2.8 trillion total parameters, and says the recipe is 4.2–12.3x more pre-training FLOPs-efficient than DeepSeek-V4-Pro-Base. The run itself is the number worth keeping: 16,000 Nvidia GB200 GPUs for 38 days across August and September, at 21% average model

Amazon's new Bee devices listen all day — for $11.99 a month

Amazon turned its always-listening Bee line into a two-device family on Thursday, and NVIDIA opened a 2.8-trillion-token Indic corpus recipe. The connecting thread: both are bets that the next AI money is in memory — yours, or your model's. Amazon put two new always-listening devices on sale, and both now lead to a subscription. Bee Frontier, the next-generation wristband that also clips to clothing, ships now at $74.99 in Onyx, Lavender and Sage — up from the $50 predecessor — with seven-day

Read more

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Brain organoids classified Braille at 83% — in a trio, not alone

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute

Amazon's new Bee devices listen all day — for $11.99 a month