Gemini 4 Neon
gemini-4-neongoogle/gemini-4-neon- π Unified 1,048,576-token context shared across input, thinking, and output
- π§ Adapted from Gemini 4 Carbon; frontier reasoning at lower latency
- π§ Four thinking levels, with HIGH default and XHIGH for deep reasoning
- ποΈ Accepts text, images, audio, video, PDFs and Office/OpenDocument files
- β‘ Long decode reaches 1M output tokens via continuation tokens
- π― Migration paths documented from Argon, Gemini 3.8 Flash and 3.1 Pro
- π Safety testing reports fewer unjustified refusals on multimodal prompts
- π Available through the Gemini API, AI Studio and Google Cloud
Google is an American multinational technology corporation and one of the world's most valuable brands. A subsidiary of parent company Alphabet Inc., Google operates across search, cloud computing, consumer electronics, and artificial intelligence. Its DeepMind and Googleβ¦
Explore 31 more models by Google βGemini 4 Neon is Google's mid-tier member of the Gemini 4 generation, released in November 2026 and positioned between Gemini 4 Carbon, which handles the hardest long-horizon projects, and Gemini 4 Hydrogen, the fast everyday execution model. Google's documentation describes Neon as balancing depth and efficiency, providing frontier-level reasoning for complex, high-frequency workflows. The DeepMind model card states Neon is adapted from Gemini 4 Carbon.
The clearest generational change versus earlier Gemini models is the token budget: Neon uses a single unified 1,048,576-token limit covering input, thinking traces, function calls and final output, rather than separate input and output limits. Synchronous responses return up to 65,536 tokens per segment, with long decode via continuation tokens extending generation to the full million.
Google's migration guidance targets users of Gemini 3.1 Pro Preview, Gemini 4 Argon and Gemini 3.8 Flash; notably, the default thinking level moves to HIGH (up from MEDIUM on Argon and 3.8 Flash), and XHIGH is newly available where those predecessors capped at HIGH. Input handling also widens, adding Microsoft Office and OpenDocument files alongside PDFs and plain text.
Capabilities include reasoning, vision and audio input, function calling and web search, with access through the Gemini API, AI Studio and Google Cloud's enterprise agent platform. Google's safety reporting describes content-safety behavior comparable to Hydrogen, with fewer unjustified refusals on multimodal prompts.
This About section is AI-generated from public sources via VeniceStats + Venice inference, with no human editing. It may contain inaccuracies.
Daily usageΒ Β·Β Gemini 4 NeonΒ Β·Β Organic providers
Top providersΒ Β·Β Gemini 4 NeonΒ Β·Β Organic
- β² Apex Ant66.2%
- Super Seeder22.0%
- Hana Gateway β 10.7%
- ZLKPro-Api1.1%
| Provider | Health | Repβ | Sales | Tokens | Routing | Input $/M | Cached $/M | Output $/M | Save |
|---|---|---|---|---|---|---|---|---|---|
| Super Seeder 0xd19fβ¦41f3 | healthy | 90.57 | $3.75 | 3.38M | #3 | $4.6876 | $0.2344 | $23.4379 | β |
| β² Apex Ant 0x73b4β¦e736 | healthy | 88.20 | $11.29 | 16.23M | #1 | $1.875 | $0.09 | $9.375 | β38% |
| Venice.ai Proxy 0x1f22β¦18c9 | healthy | 75.07 | β | β | #2 | $2.0625 | $0.1031 | $10.3125 | β31% |
| ZLKPro-Api 0x0b0bβ¦f446 | β | 37.21 | $0.19 | 551.61k | gated | $0.50 | $0.25 | $2.50 | β83% |
| klinn 0x0ee1β¦6c41 | β | 12.34 | β | β | gated | $1.8008 | $0.0868 | $9.0038 | β40% |
| Hana Gateway β 0x4ae1β¦117b | healthy | 0.51 | $1.82 | 3.17M | gated | $0.40 | $0.10 | $2.00 | β87% |
| LLMart 0x948aβ¦a000 | β | 0.00 | β | β | gated | $1.8375 | $0.0886 | $9.1875 | β39% |
"Best price" and the provider table are live Antseed catalog data (advertised $/1M tokens β or $ per generated image for unit-billed image models β not settled amounts). Reputation = buyer trust score (0-100, the Antseed SDK's own formula). "Routing" = the SDK's default buyer routing (what Antseed Desktop ships with): a trust β₯ 60 gate on the effective reputation, then cheapest-first among routable providers; live failover state (per-peer cooldowns) is buyer-side runtime and not included. Model knowledge (TLDR, provider, About) via the VeniceStats enrichment layer. Advertised catalog, not the model used in any specific purchase. "Usage on Antseed" counts only settlements whose buyers share the per-model split on-chain (metadata v2/v3, opt-in), so every usage figure is a lower bound.