Argon
A frontier AI model developed internally at Google that shows performance characteristics on the ARC-AGI-3 abstract reasoning benchmark similar to those of Astra, reflecting Google's reasoning optimization path under a multi-model parallel development strategy.
Core Facts
Gemini 4 Argon is fully available to paid users, but the Anti-Gravity feature is limited to Ultra subscribers, and Pro users cannot use it.
Argon's single output limit has expanded from 64,000 tokens to 1 million tokens.
Google DeepMind has released the next-generation frontier large model Gemini 4 Argon, positioned as a flagship model capable of sustained deep reasoning in complex, long-horizon professional workflows.
Timeline (last 90 days)
On the SEC Bench Pro benchmark, Carbon achieved 88.2%, surpassing GPT-6 Astra's 85.4% and Argon's 83.4%.
Google's internal tests show that Carbon's success rate in generating secure code is more than double Argon's.
In the Argon security team's hands-on testing, Carbon's level of writing high-quality vulnerability patches is more than ten times better than other frontier models.
Gemini 4 Carbon's unsupported claims are about half as many as Argon's, and severe behavioral deviation alerts in the internal coding environment dropped by nearly 80%.
Against indirect prompt injection attacks, Argon had a 0.7% success rate of being compromised, while Gemini 4 Carbon was not compromised even once.
Gemini 4 Carbon took the top score in 15 of 17 tests, with the models compared including Argon, GPT-6, Astra and Cloud Opus 5.5.
On the Graphworks long-horizon reasoning benchmark, Gemini 4 Carbon scored 96.2%, higher than Argon's 84.2%.
On the fifth edition of the Comprehensive Intelligence leaderboard, Gemini 4 Carbon scored 59 points, ranking first and overtaking Claude Opus 5.5, 9 points higher than Argon.
Gemini 4 Carbon achieved an 18% improvement over the previous generation Argon within two weeks.
The previous-generation model Argon scored 50 points on the Archital Analysis Intelligence Index B5 leaderboard.
32 more timeline events
All Facts (20)
Gemini 4 Argon is fully available to paid users, but the Anti-Gravity feature is limited to Ultra subscribers, and Pro users cannot use it.
90%VerifiedArgon's single output limit has expanded from 64,000 tokens to 1 million tokens.
65%VerifiedGoogle DeepMind has released the next-generation frontier large model Gemini 4 Argon, positioned as a flagship model capable of sustained deep reasoning in complex, long-horizon professional workflows.
80%Partially VerifiedThe release of Gemini 4 Carbon comes only two weeks after the release of the previous generation, Gemini 4 Argon.
90%Partially VerifiedGoogle DeepMind officially released its next-generation frontier model Gemini 4 Carbon, which it calls its "strongest model."
85%Partially VerifiedGemini 4 Carbon is priced at $2 per million input tokens and $10 for output, with a 95% discount on cached input.
80%Partially VerifiedGoogle's internal tests show that Gemini 4 Carbon's success rate in generating secure code is more than twice that of the previous generation Argon.
75%UnverifiedOn the fifth edition of the Comprehensive Intelligence leaderboard, Gemini 4 Carbon scored 59 points, ranking first and overtaking Claude Opus 5.5, 9 points higher than Argon.
60%UnverifiedIn the Argon security team's hands-on testing, Carbon's level of writing high-quality vulnerability patches is more than ten times better than other frontier models.
50%UnverifiedOn the SEC Bench Pro benchmark, Carbon achieved 88.2%, surpassing GPT-6 Astra's 85.4% and Argon's 83.4%.
50%UnverifiedGoogle's internal tests show that Carbon's success rate in generating secure code is more than double Argon's.
50%UnverifiedGemini 4 Carbon's unsupported claims are about half as many as Argon's, and severe behavioral deviation alerts in the internal coding environment dropped by nearly 80%.
50%UnverifiedGemini 4 Carbon took the top score in 15 of 17 tests, with the models compared including Argon, GPT-6, Astra and Cloud Opus 5.5.
50%UnverifiedAgainst indirect prompt injection attacks, Argon had a 0.7% success rate of being compromised, while Gemini 4 Carbon was not compromised even once.
50%UnverifiedOn the Graphworks long-horizon reasoning benchmark, Gemini 4 Carbon scored 96.2%, higher than Argon's 84.2%.
50%UnverifiedGemini 4 Carbon achieved an 18% improvement over the previous generation Argon within two weeks.
50%UnverifiedThe previous-generation model Argon scored 50 points on the Archital Analysis Intelligence Index B5 leaderboard.
50%UnverifiedArgon and Carbon are in leading positions on frontier model capabilities.
50%UnverifiedThe Argon team has made personal agents its key strategic direction for the next stage.
50%UnverifiedThe public tweet contains only one sentence, with no screenshots, product links, or full transcript of the conversation attached.
50%