Hi!
DIT Model Radar tracks new AI models daily: real models or confirmed codenames only, no tools, no hype.
Google has announced Gemini 4 Argon, a frontier model that follows the Gemini 3.8 Flash launch. It raises the output token limit from 64K to 1M and is rolling out soon to Google AI Ultra subscribers and paid API customers. Today it reaches trusted testers and cyber defenders through DeepMind's Fairwind program.

What changed: Google moved Gemini 4 from a training update to a full announcement: the model has an official name (Gemini 4 Argon), published benchmarks, a 1M-token output limit and an introductory price.
Status: Announced by Google on 2026-09-30 (official DeepMind account, Fairwind program page, 9to5Google), with specs, benchmarks and pricing published; general availability still pending
- Output token limit is 1M tokens (up from 64K), paired with expanded coding, reasoning and multimodal capabilities
- Benchmarks: DeepSWE v1.1 77.9%, CWE-bench v1 68% for vulnerability remediation (tied for first), LVBench 91.7%, AutomationBench 51.3% (#1)
- Pricing: introductory $2 per 1M input and $10 per 1M output tokens with cached input at 95% off
- Rollout: available now to trusted testers and cyber defenders via the Fairwind program, without cyber guardrails; wider release starts with Google AI Ultra subscribers and paid API customers
- Google says Argon is its most resilient model yet against indirect prompt injection, monitors its chain-of-thought for misalignment, and seals sandboxes before high-risk training runs
- Already in internal use: Argon agents freed over 300 TiB of data-centre memory and are migrating large C/C++ codebases (re2, libgav1, Fuchsia's Zircon kernel) to Rust
Gemini 4 Argon is Google's frontier release following the Gemini 3.8 Flash launch. Its benchmarks, 1M-token output limit and introductory pricing are public, but general availability is not: today it only reaches trusted testers and cyber defenders, with a broader rollout promised soon.
[ $1 LAUNCH SPECIAL ] Gemini 4 Argon on DIT 15× the tokens. Just $1. ≈12M input or ≈1.5M output tokens.* Unlocks at launch. *Illustrative from gemini-2.5-pro: $1.25/M in + $10/M out. |

Top AI models. Lower costs.
Access GPT, Claude, Gemini, and more through one OpenAI-compatible API—and save up to 70% on AI API costs. Create your DIT account, get your API key, and start building.