Models

gpt-6.2-astra — API, Pricing & Context Window | Vivgrid

Run OpenAI's gpt-6.2-astra on Vivgrid: the latest gpt-6 flagship for coding agents, with a 1M-token context window, 128K max output, and geo-distributed acceleration.

gpt-6.2-astra is the latest flagship in OpenAI's gpt-6 family, succeeding gpt-6-astra. It pairs a 1M-token context window with 128K-token max output at $8.80 per 1M input tokens and $44.00 per 1M output tokens, below gpt-6-astra's $10.00 / $50.00, for whole-repository refactors, long-running agents, and workflows where output quality outweighs token cost.

On Vivgrid, gpt-6.2-astra is served through a single OpenAI-compatible endpoint with geo-distributed acceleration across AMER, EMEA, and APAC, so requests are routed to the nearest compute region for lower latency. Moving an agent from gpt-6-astra is a model-string change in the Console, with no application code to touch.

It is available on both OpenAI surfaces — Chat Completions (/chat/completions) and the Responses API (/responses) — so coding CLIs that expect the Responses wire format, such as Codex and Pi, work against the same key.

Specifications

ProviderOpenAI
Model IDgpt-6.2-astra
Best forCoding
Available onChat Completions, Vibe coding
Context window1,000,000 tokens
Max output128,000 tokens
ModalitiesText, Image
Tool / function callingYes
Knowledge cutoff2026-05
Acceleration⚡ Geo-Distributed — AMER, EMEA, APAC

Pricing

Pricing in USD per 1M tokens.

InputCached inputOutput
$8.80$0.44$44.00

What one request costs

The rates above applied to a 20K-token prompt with a 2K-token reply — a typical single agent turn, small enough that every request is billed at the base rate.

RequestCost
Cold prompt, nothing cached$0.2640
Warm prompt, 90% of the input served from cache$0.1135

Prompt caching is billed at $11.00 per 1M cache-write tokens, with cached reads at $0.44.

Requests beyond 272K input tokens are billed at the long-context rate: $17.60 input / $0.88 cached input / $22.00 cache write / $66.00 output per 1M tokens.

Use gpt-6.2-astra in your coding agent

gpt-6.2-astra works with the same Vivgrid key and https://api.vivgrid.com/v1 endpoint as every other model in the catalog. Grab your API key from the Vivgrid Console, then drop one of the configurations below into your agent of choice.

Codex

Install Codex:

npm i -g @openai/codex

Add Vivgrid as a provider and make gpt-6.2-astra the default model in ~/.codex/config.toml. Codex talks to Vivgrid over the Responses API:

~/.codex/config.toml
model = "gpt-6.2-astra"
model_provider = "vivgrid"
preferred_auth_method = "apikey"
forced_login_method = "api"

[features]
streamable_shell = true
unified_exec = true
shell_snapshot = true

[model_providers.vivgrid]
name = "Vivgrid"
base_url = "https://api.vivgrid.com/v1"
wire_api = "responses"
experimental_bearer_token = "viv-xxxxxxxxxxxxx"

Then start Codex in your project:

codex

Already using another Vivgrid model in Codex? Switch for a single run with codex -m gpt-6.2-astra. See the full Codex tutorial.

OpenCode

OpenCode has Vivgrid built in, so you can use gpt-6.2-astra without writing any config: open /models, search for vivgrid, paste your API key and pick the model. The OpenCode tutorial walks through it step by step.

Prefer to configure it by hand? Install OpenCode:

npm i -g opencode-ai

Declare gpt-6.2-astra under the Vivgrid provider in your global OpenCode config, ~/.config/opencode/opencode.json. The @ai-sdk/openai provider talks to Vivgrid over the Responses API (/v1/responses):

~/.config/opencode/opencode.json
{
  "$schema": "https://opencode.ai/config.json",
  "model": "vivgrid/gpt-6.2-astra",
  "provider": {
    "vivgrid": {
      "npm": "@ai-sdk/openai",
      "name": "Vivgrid",
      "options": {
        "baseURL": "https://api.vivgrid.com/v1",
        "apiKey": "{env:VIVGRID_API_KEY}"
      },
      "models": {
        "gpt-6.2-astra": {
          "name": "gpt-6.2-astra",
          "tool_call": true,
          "attachment": true,
          "modalities": { "input": ["text", "image"], "output": ["text"] },
          "limit": { "context": 1000000, "output": 128000 }
        }
      }
    }
  }
}

Export your key and launch OpenCode:

export VIVGRID_API_KEY="viv-xxxxxxxxxxxxx"
opencode

Use /models to switch between gpt-6.2-astra and other Vivgrid models mid-session. See the full OpenCode tutorial.

Pi

Install Pi:

npm install -g --ignore-scripts @earendil-works/pi-coding-agent

Add gpt-6.2-astra to the Vivgrid provider in ~/.pi/agent/models.json. The provider defaults to Chat Completions, and gpt-6.2-astra overrides api at the model level to use the Responses API (/v1/responses):

~/.pi/agent/models.json
{
  "providers": {
    "vivgrid": {
      "baseUrl": "https://api.vivgrid.com/v1",
      "apiKey": "viv-xxxxxxxxxxxxx",
      "api": "openai-completions",
      "compat": {
        "sendSessionAffinityHeaders": true
      },
      "models": [
        {
          "id": "gpt-6.2-astra",
          "api": "openai-responses",
          "input": ["text", "image"],
          "contextWindow": 1000000,
          "maxTokens": 128000
        }
      ]
    }
  }
}

Make it the model Pi starts with in ~/.pi/agent/settings.json:

~/.pi/agent/settings.json
{
  "defaultProvider": "vivgrid",
  "defaultModel": "gpt-6.2-astra"
}

Then verify and start Pi:

pi --list-models vivgrid
pi

Or run a one-shot prompt without entering the TUI:

pi -p "explain the auth flow in this repo" --model vivgrid/gpt-6.2-astra

See the full Pi tutorial.

API surfaces

The Vivgrid Console catalogs gpt-6.2-astra on 2 surfaces, all behind the same API key. The endpoint shown is the one the Console previews for that surface.

SurfacePrimary endpointUse it for
Chat Completions/chat/completionsAgent projects, where Vivgrid injects the model, system prompt and tools server-side, and Model API calls that name the model per request.
Vibe coding/responsesCoding CLIs and IDE agents that drive the model themselves — point the tool at Vivgrid and keep your own loop.

Quick start

Get an API key from the Vivgrid Console, then call gpt-6.2-astra directly.

/chat/completions

curl https://api.vivgrid.com/v1/chat/completions \
  -H "Authorization: Bearer $VIVGRID_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-6.2-astra",
    "messages": [
      { "role": "user", "content": "Say hello in English, Chinese and Spanish." }
    ],
    "stream": true
  }'

/responses

curl https://api.vivgrid.com/v1/responses \
  -H "Authorization: Bearer $VIVGRID_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-6.2-astra",
    "input": "Refactor this function and explain the change.",
    "stream": true
  }'

Choosing within the gpt-6 family

Prices are USD per 1M tokens at the base rate. Rows marked this page are the model described here.

ModelContextMax outputInputOutput
gpt-6.2-astra — this page1,000,000128,000$8.80$44.00
gpt-6-astra1,050,000128,000$10.00$50.00
gpt-6.2-sol1,000,000128,000$2.20$11.00
gpt-6.1-sol1,050,000128,000$2.00$10.00
gpt-6.1-terra1,050,000128,000$1.10$5.50
gpt-6-luna1,050,000128,000$0.10$0.50

Ideal use cases

  • Frontier coding agents and CLI tools (Codex, OpenCode, Claude Code-style flows)
  • Whole-codebase reads and edits in a single 1M-token pass
  • High-stakes reasoning where answer quality outweighs token cost
  • Long agent traces that chain many tool calls across a session
  • gpt-6-astra — the prior astra release, at a higher per-token price
  • gpt-6.2-sol — the latest balanced gpt-6 model at a quarter of the price
  • claude-fable-5-2 — Anthropic's most capable model

On this page