Current instruments

Compare live Text, Code, and Agent instruments by their associated specifications, including quality, context windows, and speed.

Instruments are Specifications, not a Model

You buy a quality floor, not a model name. Each instrument is defined by measurable thresholds: Quality Score, latency, context, input length, output length, sustained throughput, uptime. Any model from any supplier that clears every threshold can fill orders for that instrument. The order book routes each request to the qualifying supplier offering the best price at that moment.

You write model: "text-prime" (or any other instrument string) in the request and call either the OpenAI Chat Completions endpoint at https://api.thegrid.ai/v1 or the Anthropic Messages beta at https://messages-beta.api.thegrid.ai/v1. The same instrument string works on both. The specific model behind any given call can change between calls. You contracted for the specification, not the model name.

At-a-glance comparison

How Quality, Context Window, Input Length, Output Length, TTFT, and Throughput are defined

Each instrument can be served by many different models, so its specs define the "lowest acceptable bar" the supplied model must meet. Some specs are floors, such as Context, Input, and Output, that models may exceed. Others are ceilings, such as TTFT, that models may come in under.

Quality Score is the minimum score a model must achieve on the associated Artificial Analysis index: Intelligence Index v4.1 for text instruments, Coding Index (current) for code instruments, and Agentic Index (current) for agent instruments. Artificial Analysis publishes a numbered v4.1 label for the Intelligence Index; Coding Index and Agentic Index are referenced as current AA indices because no separate numbered versions are exposed publicly.

Context Window represents the minimum context window, in tokens, the served model must support. Consumers of our API, this is the safe value to assume in your usage. Suppliers, feel free to support a higher value.

Input Length represents the minimum input length, in tokens, the served model must support. Consumers of our API, this is the safe value to assume in your usage. Suppliers, feel free to support a higher value.

Output Length represents the minimum output length, in tokens, the served model must support. Consumers of our API, this is the safe value to assume in your usage. Suppliers, feel free to support a higher value.

Time to First Token aka TTFT is the maximum time a supplier must serve their first token during a streaming response. This is measured as p90 within our infrastructure, so it excludes network hops to the client.

Throughput is the minimum rate at which tokens are generated after the first token, in tokens per second, during a streaming response. Also measured as p90 within our infrastructure.

Text Standard

Quality Score

≥ 24 Intelligence Index v4.1

Context Window

≥ 128K tokens

Input Length

≥ 120K tokens

Output Length

≥ 16K tokens

Time to First Token

≤ 2 seconds

Throughput

≥ 70 tok / sec

Text Prime

Quality Score

≥ 41 Intelligence Index v4.1

Context Window

≥ 196,608 tokens

Input Length

≥ 120K tokens

Output Length

≥ 30K tokens

Time to First Token

≤ 5 seconds

Throughput

≥ 40 tok / sec

Text Max

Quality Score

≥ 56 Intelligence Index v4.1

Context Window

≥ 1M tokens

Input Length

≥ 922K tokens

Output Length

≥ 128K tokens

Time to First Token

≤ 7 seconds

Throughput

≥ 30 tok / sec

Code Standard

Quality Score

≥ 30 Coding Index (current)

Context Window

≥ 128K tokens

Input Length

≥ 120K tokens

Output Length

≥ 16K tokens

Time to First Token

≤ 3 seconds

Throughput

≥ 60 tok / sec

Code Prime

Quality Score

≥ 57 Coding Index (current)

Context Window

≥ 196,608 tokens

Input Length

≥ 120K tokens

Output Length

≥ 30K tokens

Time to First Token

≤ 6 seconds

Throughput

≥ 40 tok / sec

Code Max

Quality Score

≥ 74 Coding Index (current)

Context Window

≥ 1M tokens

Input Length

≥ 922K tokens

Output Length

≥ 128K tokens

Time to First Token

≤ 9 seconds

Throughput

≥ 30 tok / sec

Agent Standard

Quality Score

≥ 13 Agentic Index (current)

Context Window

≥ 128K tokens

Input Length

≥ 120K tokens

Output Length

≥ 16K tokens

Time to First Token

≤ 3 seconds

Throughput

≥ 60 tok / sec

Agent Prime

Quality Score

≥ 35 Agentic Index (current)

Context Window

≥ 196,608 tokens

Input Length

≥ 120K tokens

Output Length

≥ 30K tokens

Time to First Token

≤ 6 seconds

Throughput

≥ 40 tok / sec

Agent Max

Quality Score

≥ 47 Agentic Index (current)

Context Window

≥ 1M tokens

Input Length

≥ 922K tokens

Output Length

≥ 128K tokens

Time to First Token

≤ 8 seconds

Throughput

≥ 30 tok / sec

To call each of these instruments via the Consumption API, you pass the instrument string in the model parameter, for example text-prime. The order book routes the request to a qualifying supplier offering the best price.

Text instruments

Three tiers covering general-purpose text generation, reasoning, summarization, and retrieval-augmented work. Text instruments are live today.

Text Prime (default)

• Production default for everyday text generation. Strong reasoning at a fraction of frontier cost. Most workloads belong here.
• Use for: RAG and retrieval-augmented apps, content drafting, summarization, customer-facing generation where quality matters but frontier reasoning is overkill.
• Not sure where to start? Start on text-prime and move up or down from there.

Text Max

• For high-stakes work where errors compound. The 1M context window handles long-context synthesis across many documents in a single pass.
• Use for: legal review, clinical reasoning, financial analysis, security incident triage.
• Route by exception, not by default. If you do not need the full context window or the highest Quality Score floor, you are paying for headroom that goes unused.

Text Standard

• For high-volume, low-stakes work where speed and unit economics matter more than reasoning depth.
• Use for: classification and routing, chatbot first-touch, batch summarization, pipeline glue.
• Latency advantage. The 2 second TTFT cap keeps interactive responses feeling instant; the lower Quality Score floor keeps per-call costs manageable at volume.
• Tricky calls? Send them up to Prime when they hit.

Code instruments

Three tiers covering software engineering work, evaluated on real coding tasks, not chat.

Code Prime (default)

• Production default for daily coding work. Fast enough for interactive use, smart enough for non-trivial tasks. Most production coding agents land here.
• Use for: code completion in your IDE, "write this function" requests, automated PR review and feedback, standard debugging on stack traces, feature implementation where correctness is binary (the tests pass or they do not).

Code Max

• For changes that span the whole codebase. The 1M context window and higher Coding Index floor handle codebases that overwhelm smaller models.
• Use for: full-repo analysis, cross-repo migrations across multiple services and API boundaries, complex multi-layer debugging (race conditions, distributed system failures, subtle type mismatches), multi-file refactors touching thousands of lines.

Code Standard

• For autocomplete, linting, and batch edits. High frequency, well-defined tasks where TTFT under 3 seconds and high throughput dominate.
• Use for: inline suggestions firing on every keystroke, automated style enforcement, rename-in-200-files refactors, API call signature updates, boilerplate generation.
• Reasoning depth is not the constraint here.

Agent instruments

Three tiers covering autonomous and multi-step tool use, benchmarked on real agentic work.

Agent Prime (default)

• Production default for daily agent work.
• Use for: production agent loops running all day (customer service, data processing, monitoring), multi-step tool chains of 5 to 20 calls (database lookup, response formatting, ticket update), chain-of-thought reasoning workflows where the agent thinks through options before acting.
• Common pattern: pair agent-prime for planning and tool use with code-prime for the actual code-writing.

Agent Max

• For autonomous work measured in hours, not minutes. The higher Agentic Index floor and long-horizon validation matter more than per-call cost at this tier.
• Use for: independent research that reads papers, searches the web, and produces a structured report after hours of unattended work; deep tool chains of 50+ calls where each step depends on the last; long-horizon infrastructure management, data pipeline construction, high-stakes automation acting on production systems.

Agent Standard

• For fast tool calls and orchestration glue. The lightweight layer in an agent system that parses responses, decides which tool to call next, and formats inputs (these calls fire hundreds of times per agent run).
• Use for: single-purpose agents that do one thing repeatedly, high-throughput orchestration where hundreds or thousands of agent instances run in parallel, routing and triage agents that decide whether a request handles directly or escalates.

Lab latest markets

Lab latest markets are for buyers who want a model-family route rather than a task-tier specification. You still pass the market string in the model field. The underlying model and supplier can change as newer routes qualify.

Current lab latest instruments are gpt-sol-latest, claude-opus-latest, gemini-pro-latest, minimax-latest, glm-latest, deepseek-pro-latest, kimi-latest, and bytedance-pro-latest.

Use these current limits when configuring tools that require explicit context, input, or output values:

Instrument	Context window	Maximum input	Maximum output
gpt-sol-latest	1,000,000	922,000	128,000
claude-opus-latest	1,000,000	922,000	128,000
gemini-pro-latest	1,000,000	922,000	65,536
minimax-latest	1,000,000	922,000	512,000
glm-latest	1,000,000	922,000	128,000
deepseek-pro-latest	1,000,000	922,000	384,000
kimi-latest	262,144	256,000	262,144
bytedance-pro-latest	262,144	256,000	131,072

These values can change as Lab Latest routes evolve. Check /v1/models or the dashboard for current metadata before increasing configured limits.

How to choose

• The biggest savings come from routing across tiers, not from picking one.
• Common production setup: Standard for triage and orchestration glue, Prime for the calls that actually need to reason, Max only when the context is huge or the cost of an error is high.
• Mix task types freely. An agent on agent-prime for planning, code-prime for implementation, text-standard for the run summary.

For more details on how we revise our specifications and qualifying criteria for suppliers, see How instrument specifications evolve.