Gemini 4 Neon API
Gemini 4 Neon is the middle model of the Gemini 4 line: it reasons more deeply than Hydrogen and finishes tasks faster and for less than Carbon, aimed at complex coding, analytical reasoning and production agent workflows. Google released it on 2026-11-18 and began rolling it out the same day on the paid tier of the Gemini API, in AI Studio, in the Gemini app (Pro and Ultra) and in Antigravity (paid Pro and Ultra); in the Gemini app’s model picker it takes Argon’s place. It lists at $3 input / $15 output per million tokens, with no introductory discount and no free tier. Per token it costs more than Carbon’s promotional price until 2026-11-30; per task, Google says it is 32% cheaper than Carbon at the default level. This page covers the official benchmarks, how it compares with Argon, Carbon, Hydrogen, Claude Opus 5.5, Fable 5.2 and GPT-6.2 Sol, which thinking level to pick, and the parameters and limits when you call it through Wokey.
Gemini 4 Neon at a glance
- Release
- Released on 2026-11-18 with a gradual rollout: the paid tier of the Gemini API (AI Studio, Android Studio) and Gemini for business; the Gemini app for Google AI Pro and Ultra (“Deeper analysis” in the model picker); the Antigravity desktop app, CLI and IDE extensions for paid Pro and Ultra; GitHub Copilot on launch day
- Positioning
- Google says it “sits between Gemini 4 Hydrogen and Gemini 4 Carbon”, balancing depth and efficiency: Carbon stays the pick for the hardest architecture, debugging, science and security problems, and Hydrogen the default for everyday coding and agentic workflows. Google says it costs 32% less per task than Carbon at the default high level; Antigravity’s Orchestrator (Neon) has Neon plan and Hydrogen execute
- Context and output
- 1,048,576-token context window in the Gemini API, shared by input, thinking and output; up to 65,536 output tokens per synchronous request, about 1M with background long decode. The “up to 512K” in Antigravity’s model table is where the Antigravity agent compacts its conversation, not the model’s limit. Accepts text, images, audio, video and PDF and outputs text; knowledge cutoff March 2026
- Thinking levels
- Four levels, low / medium / high / xhigh, with high as the API default, as on Hydrogen; xhigh is for hard multi-step engineering, analytical and scientific problems where high falls short. minimal returns an error. Thinking tokens are billed at the output rate and max_output_tokens caps thinking and output together; hard prompts can run for several minutes, and Google recommends client timeouts of at least 15 minutes
- Official pricing
- $3 input / $15 output per million tokens, $0.15 cache read; no introductory price, no long-context surcharge and no free tier. Batch and Flex are half price ($1.50 / $7.50) and Priority is 1.8x. Implicit caching only (prompts of at least 4,096 tokens, hits billed at 5% of the input rate). For comparison, Hydrogen is $1.50 / $7.50; Carbon and Argon are $2 / $10 through 2026-11-30 and $4 / $20 from 12-01
- API features
- Function calling, structured outputs, code execution, Google Search and Maps grounding, URL context, file search, computer use, Batch, Flex and Priority; no Live API, image or audio generation, or explicit caching. temperature, top_p and top_k are ignored
- Subscriptions
- Antigravity’s model table: not on free or AI Plus; Google AI Pro needs a paid plan (or a trial bundled with an eligible premium device), and Ultra and Enterprise include it; Neon is now the default model for new Antigravity conversations. Orchestrator (Neon) is on Pro and Ultra, Orchestrator (Carbon) on Ultra only. Neon shares the “Gemini Models” quota with Hydrogen and 3.8 Flash, which is drawn down by token cost, so Neon uses it up faster
- Model ID
- gemini-4-neon (the only stable version so far)
· Sources: Model card · Gemini API · Model · Gemini API · What’s new · Gemini API · Pricing · Google · Antigravity · Artificial Analysis
Official benchmarks: Gemini 4 Neon vs Claude Opus 5.5 vs GPT-6.2 Sol
From Google DeepMind’s Gemini 4 Neon model card (2026-11-18). These are vendor-run results, Google also supplied the competitor scores, and the column headers do not state each model’s thinking level; “—” means Google listed no score.
| Benchmark | Gemini 4 Neon | Claude Opus 5.5 | GPT-6.2 Sol |
|---|---|---|---|
| Vals Index v2.2 | 70.9% | 70.1% | 66.1% |
| CursorBench 4.1 | 65.4% | 58.1% | — |
| Terminal-Bench 4.1 | 65.2% | 65.8% | 61.5% |
| FrontierCode 2.0 | 46.2% | 47.3% | 43.8% |
| SWE-sweep (provider harness) | 39.0% | 10.1% | 8.3% |
| SWE-sweep (mini-SWE harness) | 12.8% | 3.6% | 11.6% |
| AutomationBench | 56.3% | 42.5% | 41.3% |
| Agents’ Last Exam v1.1 | 75.8% | 71.6% | 73.0% |
| RiemannBench | 84.0% | 80.0% | 80.0% |
| Terminal-Bench-Science 0.1 | 70.5% | 63.3% | 68.0% |
| HLE-Diamond | 71.3% | 67.9% | 64.0% |
| MKA (Climb WSR) | 56.5% | 53.0% | 52.7% |
| Chartography | 73.5% | 66.3% | 75.1% |
| LVBench | 89.0% | 83.7% | 87.6% |
| GraphWalks (256K–1M, BFS F1) | 95.2% | 66.8% | 80.1% |
| CWE-Bench v2 | 70.7% | 67.1% | 66.0% |
| SEC-bench Pro | 90.5% | — | 88.3% |
| WebExploitBench v1.2 | 51.3% | 37.1% | — |
Google’s table has 19 rows and Neon leads 16 of them. Claude Opus 5.5 is slightly higher on Terminal-Bench 4.1 (65.8% vs 65.2%) and FrontierCode 2.0 (47.3% vs 46.2%), and GPT-6.2 Sol is higher on Chartography (75.1% vs 73.5%). The row not shown above is the Artificial Analysis Intelligence Index v5.0.1: 61 for Neon, 59 for Opus 5.5 and 57 for Sol. In the SWE-sweep provider-harness row Neon ran in Google’s own Antigravity; on the shared mini-SWE harness it scores 12.8%. The model card has no Carbon or Argon column.
Which to use: Gemini 4 Neon, Argon, Carbon, Hydrogen, Claude or GPT-6.2
Versus Gemini 4 Argon: has it been replaced
In the Gemini app’s model picker Neon has taken Argon’s place; the options are now Auto, Everyday help (Hydrogen), Deeper analysis (Neon) and Code & complex work (Carbon). Argon remains available in the Gemini API and Antigravity, and Google has published no retirement date for it. Neon is clearly stronger: in Arena’s Agent Arena Neon (high) ranks 2nd with 1336 points and Argon (high) 11th with 1229, and on the Artificial Analysis Intelligence Index Neon scores 60 at high against Argon’s 50. Argon costs $2 / $10 now and $4 / $20 from 12-01, while Neon stays at $3 / $15, so from December Neon is 25% cheaper. Both take the same request format, so switching only changes the ID; Argon defaults to medium and stops at high, while Neon defaults to high and adds xhigh.
Versus Gemini 4 Carbon: more per token, less per task
Per token, Neon’s $3 / $15 is 50% above Carbon’s promotional $2 / $10 until 2026-11-30; once Carbon returns to $4 / $20 on 12-01, Neon is 25% cheaper. Per task, Google says Neon costs 32% less than Carbon at the default high level. Artificial Analysis measured Neon at 61 on its Intelligence Index at xhigh, 1 point above Carbon (high), at $6.44 against $7.76 per task (Carbon at standard pricing), about 17% less, and in about 21 minutes against 36; at high it scores 60, level with Carbon, for $3.74 per task, under half of Carbon’s cost. Neon outputs about 110 tokens per second against about 57 for Carbon. In Antigravity Carbon is Ultra-only, while paid Pro also gets Neon. Google’s advice is to keep Carbon for the hardest architecture, debugging, science and security problems and to try Neon first for other complex work.
Versus Hydrogen, and how Orchestrator pairs them
Hydrogen costs half as much per token ($1.50 / $7.50) and Google keeps it as the default for everyday coding and agentic workflows; Neon reasons more deeply and suits complex coding, analytical reasoning and long workflows. On the Artificial Analysis Intelligence Index Hydrogen tops out at 54 (xhigh) against Neon’s 61, and Hydrogen (high) ranks 9th in Agent Arena. Vals found Neon needs fewer steps: on Code Migration it averaged 160 turns per task against 234 for Carbon and 574 for Hydrogen, although Neon scored below both there. Antigravity’s Orchestrator (Neon) has Neon plan and Hydrogen execute; its configuration is fixed, the thinking level cannot be changed, and it exists only in Antigravity, with no matching API model. Both models share the same thinking levels, defaulting to high and supporting xhigh.
Versus Claude Opus 5.5, Fable 5.2 and GPT-6.2
Google compared Neon with Opus 5.5 and GPT-6.2 Sol at launch, and Neon leads 16 of the 19 rows; terminal coding is close (65.2% vs 65.8% on Terminal-Bench 4.1). List prices are $4 / $20 for Opus 5.5, $8 / $40 for Fable 5.2 and GPT-6.2 Astra, $2 / $10 for GPT-6.2 Sol and $3 / $15 for Neon. On the Artificial Analysis Intelligence Index Neon (xhigh) scores 61 and ranks 3rd, behind GPT-6.2 Astra (max) and Claude Fable 5.2 (max), at $6.44 per task, about 1.6x Astra. On the Coding Agent Index, Neon (xhigh) in Antigravity CLI scores 59, while Fable 5.2 (xhigh) and Opus 5.5 (max) in Claude Code both score 58, at $6.86 per task against $15.47 and $12.61. On the Vals Index Neon scores 70.9%, 0.4 points behind Fable 5.2, at less than half its cost per task. Early hands-on reports are mixed, and one tester found Opus 5.5 did better on the same task at about twice Neon’s cost.
Which thinking level: high or xhigh
Google recommends the default high for most tasks and xhigh only for hard multi-step problems where high falls short. Artificial Analysis measured all four levels: low scores 44, medium 55, high 60 at $3.74 per task and xhigh 61 at $6.44; xhigh takes about 57 seconds to the first answer token on average, high about 41. Vals found high keeps 98% of the xhigh score at roughly half the cost and 40% less time, with more direct answers. Use high for everyday complex work, drop to medium for interactive use, and turn on xhigh only when high cannot solve the problem. Through Wokey, choose the level with reasoning_effort; it runs at medium when you omit it, so pass high explicitly to match the Google API default.
Independent evaluations and official figures
- Artificial Analysis: On Intelligence Index v5.0.1 Neon scores 61 at xhigh and ranks 3rd, behind the top settings of GPT-6.2 Astra and Claude Fable 5.2, 1 point above Gemini 4 Carbon (high); at high it scores 60. At xhigh it costs $6.44 and takes about 21 minutes per task, about 17% and 40% less than Carbon, while using about 15% more output tokens; at high it costs $3.74 per task. Running the full index at xhigh took 283M output tokens (median 65.1M), which makes it a very verbose model; output speed is about 110 tokens per second, with 56.5 seconds to the first answer token at xhigh. On the Coding Agent Index (Antigravity CLI, xhigh) it scores 59, in line with Fable 5.2 and Opus 5.5 in Claude Code (58 each), at $6.86 per task. Artificial Analysis calls it “amongst the leading models in intelligence, but somewhat expensive when comparing to other models of similar price”, and “notably slow per task and very verbose”. Every Wokey meter is 30% of Google’s list price, so the same task comes to about $1.12 at high and $1.93 at xhigh.
- Google launch post: Google says Neon costs 32% less per task than Carbon at the default high level, 50% to 80% less than top competing frontier models, and 36%+ less than competitors at their highest setting when run at xhigh. It ranks 1st on the Artificial Analysis Coding Agent Index (58.6), 2nd on the Vals Index (70.9%), 2nd in Agent Arena (1336) and 1st on Vending-Bench 2 ($16,843), and scores 70.7% on CWE-bench v2, 3.4 points above Carbon; SWE-sweep is 12.8% on the mini-SWE harness and 39.0% in Antigravity. These are Google’s own figures.
- Arena: In Agent Arena (4,200+ real-world agent sessions) Neon (high) ranks 2nd with 1336 points, Google’s highest-ranked model, ahead of Carbon (high, 4th, 1303), Hydrogen (high, 9th) and Argon (high, 11th, 1229). Its median cost per task is $2.82: it scores above 3rd-place Claude Opus 5.5 (max) for 53% less per task and costs 69% less than 1st-place Claude Fable 5.2 (max).
- Vals AI: Neon scores 70.9% on the Vals Index, 2nd overall and 0.4 points behind Claude Fable 5.2 at less than half its cost per task, and ranks 1st on the Finance and Legal indexes, Tax Agent and Legal Research. At high effort it scores 70.5%, keeping 98% of the score at roughly half the cost and 40% less time, 6th overall.
Calling Gemini 4 Neon through Wokey: parameters and limits
This is how the Wokey gateway handles gemini-4-neon. Coming from Argon, Carbon or Hydrogen, check the thinking level, context length and max_tokens besides the model ID.
- The interface is OpenAI-compatible /v1/chat/completions: set the base URL to https://api.wokey.ai/v1, use your Wokey API key, and set model to gemini-4-neon. opencode, OpenClaw, Hermes Agent and other OpenAI-compatible clients are configured the same way.
- reasoning_effort values low, medium, high and xhigh map directly to Neon’s four thinking levels; without it, or with any other value such as minimal or max, the request runs at medium. Google’s own API defaults to high, so pass "high" explicitly to match it.
- Context: Wokey accepts up to 1,048,576 tokens of input and output combined for gemini-4-neon. The 512K in Antigravity’s model table is where Antigravity’s own agent compacts its conversation history, and API requests are not cut off there.
- Without max_tokens the limit defaults to 65,536, which is also Wokey’s current output ceiling; long decode and background mode are not supported yet. Thinking tokens count toward max_tokens and are billed at the output rate, so do not carry over a hard-coded low cap such as 8,192; to spend fewer tokens, lower the thinking level instead.
- Neon thinks for a long time: Artificial Analysis measured about 57 seconds to the first answer token at xhigh and about 41 at high, and Google says hard prompts can take several minutes. Use stream: true and set the client timeout to 15 minutes or more.
- Google ignores temperature, top_p and top_k, so you can leave them out.
- Text and image input are accepted today; audio, video and PDF input are not yet open on Wokey.
- Tool calling: declare tools in the tools field instead of prompting the model to print JSON tool calls, which Google says can return MALFORMED_FUNCTION_CALL. The gateway stores the thought signatures Neon returns, so send back tool_calls and tool results as usual on the next turn.
- Structured outputs: use response_format with json_schema, and keep the schema within the OpenAPI subset Gemini accepts (type, properties, required, items, enum, description, and nullable via anyOf with null). json_object, or a schema with $ref, oneOf or pattern, currently returns 400 structured_output_supply_unavailable.
- Caching is Google’s implicit caching, with no cache_control; keep long sessions on a stable prefix to hit it. Cache reads cost 5% of the input rate.
Calling Gemini 4 Neon through Wokey (/v1/chat/completions)
curl https://api.wokey.ai/v1/chat/completions \
-H "Authorization: Bearer $WOKEY_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gemini-4-neon",
"reasoning_effort": "high",
"stream": true,
"max_tokens": 65536,
"messages": [{ "role": "user", "content": "Plan and implement the migration of this service to the new API" }]
}'- Model ID
gemini-4-neon- Vendors
- google, devin
- Supply status
- Currently callable
Catalog pricing snapshot
Prices are per 1M tokens; request-time pricing controls settlement.
| Meter | Wokey | Official reference | Savings |
|---|---|---|---|
| Input | $0.9 | $3 | 70% |
| Output | $4.5 | $15 | 70% |
| Cache read | $0.045 | $0.15 | 70% |
| Cache write | — | — | — |
Official price source: https://aistudio.google.com/docs/pricing
Model capabilities
- Context
- 1,048,576 tokens
- Max output
- 65,536 tokens
- Streaming
- Supported
- Tools
- Supported
- Vision
- Supported
- Supported API forms
/v1/chat/completions
Send a request
/v1/chat/completions
curl https://api.wokey.ai/v1/chat/completions \
-H "Authorization: Bearer $WOKEY_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"gemini-4-neon","messages":[{"role":"user","content":"Hello"}]}'Use Gemini 4 Neon in your tools
The gateway is https://api.wokey.ai (OpenAI-compatible clients usually take https://api.wokey.ai/v1), and the model ID is gemini-4-neon. Open the guide for the client you use.
- Use Gemini 4 Neon in opencode: OpenAI Chat Completions · base URL and API key setup
- Use Gemini 4 Neon in DeepSeek Harness: OpenAI Chat Completions · base URL and API key setup
- Use Gemini 4 Neon in Cherry Studio: OpenAI Chat Completions · base URL and API key setup
- Use Gemini 4 Neon in WorkBuddy: OpenAI Chat Completions · base URL and API key setup
- Use Gemini 4 Neon in OpenClaw: OpenAI Chat Completions · base URL and API key setup
- Use Gemini 4 Neon in Hermes Agent: OpenAI Chat Completions · base URL and API key setup
- Use Gemini 4 Neon in Magpie: OpenAI Chat Completions · base URL and API key setup
- OpenRouter alternative: Wokey vs OpenRouter: Compare token prices, payment fees, API forms, and upstream sources.
- Verifiable AI API: which models and API forms carry response proofs: Response proofs currently cover only raw passthrough Claude Messages and GPT Responses on official routes.
Frequently asked questions
Is Gemini 4 Neon the same as Gemini 4 Pro, and has it replaced Argon?
Google has not released a model called “Gemini 4 Pro”. The Gemini 4 line is named after elements: Carbon is the strongest, Neon sits between Carbon and Hydrogen, Hydrogen takes the place Flash used to hold, and Argon is a leaner reasoning model that uses fewer tokens than Carbon. The API model ID is gemini-4-neon, and there is no gemini-4-pro ID. In the Gemini app Neon has taken Argon’s place, and the model picker now offers Auto, Everyday help (Hydrogen), Deeper analysis (Neon) and Code & complex work (Carbon); Argon remains available in the Gemini API and Antigravity, and Google has published no retirement date for it.
When was Gemini 4 Neon released, can I use it for free, and which plans include it?
It was released on 2026-11-18 and began rolling out the same day. There is no free way to use it: the Gemini API offers it on the paid tier only; the Gemini app offers it to Google AI Pro and Ultra subscribers; Antigravity offers it on paid Pro and Ultra, not on free or AI Plus, and the only Pro trials that qualify are those bundled with eligible premium devices, while users report other promotional Pro plans (such as carrier bundles) show it greyed out. Business users can get it through Gemini for business, and GitHub Copilot added it on launch day.
How much does the Gemini 4 Neon API cost, and why does it cost more per token than Carbon?
It lists at $3 input / $15 output per million tokens with $0.15 cache reads; Batch and Flex are half price ($1.50 / $7.50), Priority is 1.8x, the rate is flat across the 1M context window, and there is no introductory price. That is twice Hydrogen’s price. Against Carbon and Argon it depends on the date: through 2026-11-30 they are on a promotional $2 / $10, so Neon costs 50% more per token; from 12-01 they return to $4 / $20 and Neon is 25% cheaper. Per task, Google says Neon costs 32% less than Carbon at the default level, and Artificial Analysis measured about 17% less at xhigh. Through Wokey it costs $0.90 input / $4.50 output per million tokens, 70% below Google’s list price.
Gemini 4 Neon vs Argon vs Carbon vs Hydrogen: what is the difference, and which should I use?
Carbon is the strongest and handles the hardest architecture, debugging, science and security problems; Neon comes next, faster and cheaper per task, for complex coding, analytical reasoning and high-frequency production work; Hydrogen is the fastest and cheapest, for everyday coding and subagents. On the Artificial Analysis Intelligence Index Neon scores 61 at xhigh and 60 at high, Carbon 60, Hydrogen up to 54 and Argon 50; in Agent Arena Neon ranks 2nd, Carbon 4th, Hydrogen 9th and Argon 11th. Hydrogen lists at $1.50 / $7.50, Neon at $3 / $15, and Carbon and Argon at $2 / $10 now and $4 / $20 from 2026-12-01. Between Argon and Neon, new projects can go straight to Neon; in Antigravity, Orchestrator (Neon) has Neon plan and Hydrogen execute.
Gemini 4 Neon vs Claude Opus 5.5, Fable 5.2 and GPT-6.2 Sol: which is better?
List prices are $3 / $15 for Neon, $4 / $20 for Opus 5.5, $8 / $40 for Fable 5.2 and GPT-6.2 Astra, and $2 / $10 for GPT-6.2 Sol. Neon leads 16 of the 19 rows in Google’s model card, with Opus 5.5 slightly ahead on Terminal-Bench 4.1 and FrontierCode 2.0. Independently, Neon (xhigh) scores 61 on the Artificial Analysis Intelligence Index, below the top settings of Astra and Fable 5.2 and above Opus 5.5; on the Coding Agent Index Neon scores 59 against 58 for both Fable 5.2 and Opus 5.5, at about half Opus 5.5’s cost per task; on the Vals Index it is 0.4 points behind Fable 5.2 at less than half the cost per task. Choose Fable 5.2 or Astra when the ceiling matters most and try Neon first for value; some early hands-on reports still find Opus 5.5 better on specific tasks.
Should I use high or xhigh thinking with Gemini 4 Neon?
Neon has four levels, low, medium, high and xhigh, and the Google API defaults to high. Google recommends high for most tasks and xhigh only for hard multi-step problems where high falls short. Artificial Analysis measured just 1 more point at xhigh than at high (61 vs 60), with the cost per task rising from $3.74 to $6.44, and Vals found high keeps 98% of the score at roughly half the cost and 40% less time. Medium scores 55 and low 44. Through Wokey, reasoning_effort accepts all four values and runs at medium when omitted.
Is the Gemini 4 Neon context window 1M or 512K, and how much can it output?
It is 1M. In the Gemini API the window is 1,048,576 tokens, shared by input, thinking and output; a synchronous request outputs up to 65,536 tokens by default, and background long decode reaches about 1M. The “up to 512K” in Antigravity’s model table is the point at which the Antigravity agent compacts its conversation into a summary: Antigravity sets it at 512K for every Gemini 4 model and at 256K for Claude, although those models accept 1M. Through Wokey you can send up to 1,048,576 tokens, with up to 65,536 output tokens per request and no long decode yet.
How do I use Gemini 4 Neon in opencode and other coding tools without a Google subscription?
Antigravity signs in with a Google account and draws on that plan’s quota rather than an API key, and it is not available on free or AI Plus. Without a subscription you have two options: a paid-tier Gemini API key (billed at $3 / $15, with no free tier), or a pay-as-you-go API such as Wokey with model gemini-4-neon and base URL https://api.wokey.ai/v1, which works in opencode, OpenClaw, Hermes Agent and other OpenAI-compatible clients. GitHub Copilot has also added it in the Copilot app, CLI and VS Code.
How much does the Gemini 4 Neon API cost, and how much does it save versus the official API?
Wokey lists input at $0.9 per 1M tokens and output at $4.5 per 1M tokens; the official references are $3 and $15. That is 70% lower for input and 70% lower for output. Request-time pricing controls settlement, and the vendor pricing source is linked on this page.
What is the Gemini 4 Neon API model ID?
The model ID for Gemini 4 Neon on Wokey is gemini-4-neon. Use it exactly as written in the request body's model field or in your client's model setting (for example ANTHROPIC_MODEL in Claude Code, or model in the Codex CLI config.toml); GET https://api.wokey.ai/v1/models also lists it.
Which API forms does Gemini 4 Neon support, and how do I call it through Wokey?
Gemini 4 Neon supports /v1/chat/completions (Chat Completions requests). Set the API base URL to https://api.wokey.ai, authenticate with a Wokey API key, and use the canonical model ID gemini-4-neon; you can start with /v1/chat/completions. The Send a request section includes a curl example for every supported API form.
How do I use Gemini 4 Neon in opencode, Hermes Agent, and similar clients?
These clients call Gemini 4 Neon over OpenAI Chat Completions: point the API address at https://api.wokey.ai (clients that expect an OpenAI SDK-style address take https://api.wokey.ai/v1), add your Wokey API key, and use the model ID gemini-4-neon. Each client's guide shows its config format.
What are the context window and maximum output for Gemini 4 Neon?
The catalog context window is 1,048,576 tokens and the maximum output is 65,536 tokens. These are separate limits, and a request must also satisfy the upstream constraints in effect when it is processed.
How are cache reads and cache writes priced for Gemini 4 Neon?
Cache read: Wokey lists $0.045 per 1M tokens and the official reference is $0.15 per 1M tokens. Cache write: the vendor does not publish a separate meter, so Wokey also shows — as the public price.
How does calling Gemini 4 Neon through Wokey differ from OpenRouter?
Wokey lists Gemini 4 Neon at $0.9 input / $4.5 output per 1M tokens, against an official reference of $3 / $15. OpenRouter states it adds no markup on inference and charges its fee when you buy credits (5.5% by card). Wokey carries a curated set of models and has no per-provider routing parameters, so OpenRouter fits better if you need a wider catalog. The OpenRouter alternative page has the full comparison.
How do Gemini 4 Neon and Gemini 4 Argon differ in price and context?
In the published price comparison, Gemini 4 Neon lists input/output at $0.9 / $4.5 with a 1,048,576-token context window; Gemini 4 Argon lists $1.2 / $6 with a 1,048,576-token context window. This is a factual price-and-capacity comparison, not a quality ranking; also compare their supported API forms and capabilities.
Related models and documentation
Get API key · gemini-4-neon