Grok 4.3 ranks 55th of 186 tracked models on Artificial Analysis's Intelligence Index, a middling score that undersells its real strength: a large GDPval-AA agentic-benchmark gain over its predecessor. It suits xAI-ecosystem teams valuing long context and agentic tool use over a broad benchmark sweep, especially those already on AWS Bedrock or Azure.
Grok 4.3 is xAI's flagship reasoning model, released around April 2026 with a 1 million token context window for text and image input. Its clearest verified improvement over predecessor Grok 4.20 comes from a large GDPval-AA agentic-benchmark gain, though xAI has not published an official model card for this release.
Where it sits
- $1.56/M$ per 1M tokensBlended price (3:1)Lower is better#30 / 64peer median $1.70/Mvendor price, checked by HokAI
- 94 tok/stokens/sOutput speedHigher is better#17 / 39peer median 90 tok/scited: Artificial Analysis
- --% solvedSWE-bench VerifiedHigher is better-- / 28peer median 78.3%per source, see benchmark scores
- --% correctGPQA DiamondHigher is better-- / 44peer median 88.3%per source, see benchmark scores
Priced around the middle of the 64 GA models with a published price (rank 30), and rank 17 of 39 on output speed as cited from Artificial Analysis.
Ranks are against GA models on HokAI that publish the same figure; ties share a rank.
Provider: xAI · Family: Grok 4
Context window: 1,000,000 tokens
Input modalities: text, image · Output: text
About Grok 4.3
Grok 4.3 is xAI's mid-2026 flagship model for chat and reasoning tasks, reaching the direct API around April 17, 2026 in a quiet rollout before Artificial Analysis dated its formal release to April 30, 2026. It sits between Grok 4.20 (predecessor, launched February to March 2026) and Grok 4.5 (successor, launched July 8, 2026) in xAI's lineup, though Grok 4.5 is positioned as a separate, cheaper coding-focused sibling rather than a strict replacement. xAI has not disclosed Grok 4.3's parameter count, and no source confirms if it runs a dense architecture or a mixture-of-experts design; no official model card has been published for this version, a gap that persists from earlier Grok releases and that Fortune previously flagged as a recurring pattern for the product line.
On Artificial Analysis's current live tracking, Grok 4.3 scores 38 on the Intelligence Index, ranked 55th of 186 tracked models, with an output speed of 94.2 tokens per second and a time-to-first-token around 22 seconds, consistent with a model that reasons before producing its first output token. Artificial Analysis's April 30, 2026 launch article separately reported a higher Intelligence Index of 53 at release, a discrepancy from the current live score that likely reflects a reasoning-effort tier difference or a later recalibration; both figures come from Artificial Analysis rather than an independent cross-check. The clearest, most consistent benchmark result is on GDPval-AA: Grok 4.3 scored an Elo of 1500, a 321-point jump from Grok 4.20's 1179, though it still trails GPT-5.5 by 276 Elo points with roughly a 17% expected win rate in that specific evaluation. Other reported gains include tau-squared-Bench Telecom at 98% and AA-Omniscience accuracy up 8 points, though the model's non-hallucination rate on that same benchmark dropped 8 points, a real tradeoff. xAI has not published GPQA Diamond, AIME 2025, MMLU-Pro, SWE-bench Verified, ARC-AGI 2, or LMArena scores specifically for Grok 4.3 in any source found, so this page does not report them.
The model carries a 1 million token context window, matching Grok 4.20's single-model variants and confirmed directly from xAI's own developer docs; no separate maximum output token limit is documented. Modalities are text and image input with text-only output, per xAI's own docs page, contradicting some secondary blog coverage claiming native video input that could not be verified against the primary documentation. Reasoning is configurable, and the model supports native function calling and structured outputs per xAI's developer documentation. Pricing at the API level for smaller prompts is unchanged from Grok 4.20's published rate, with a substantial discount for cached input and an additional discount on the batch API; xAI's docs also list a large-prompt tier that doubles both input and output pricing, previously unreflected on this page and detailed in the pricing FAQ below. Training data cutoff date could not be reliably confirmed: xAI's docs table shows a date that appears identical across every model row, a strong signal it is a stale placeholder rather than a genuine per-model figure, so this page leaves the cutoff unlisted rather than publish an unverified date.
Grok 4.3 is reachable through the direct xAI API (model IDs grok-4.3, grok-4.3-latest, and the grok-latest alias) across the us-east-1, eu-west-1, and us-west-2 regions, and reached AWS Bedrock around June 2026 through a Mantle inference engine at an OpenAI-compatible endpoint. Microsoft Azure AI Foundry also lists Grok 4.3 with its own default Content Safety layer, and OpenRouter and Vercel AI Gateway offer the model as well; availability on Cloudflare, Snowflake, or Databricks Mosaic specifically for this version was not independently confirmed. One user-reported issue on AWS re:Post describes a slow vision encoder when running Grok 4.3 on Bedrock Mantle, though the full workaround thread could not be independently verified. The model is closed-weight and proprietary, API-only, with no license for self-hosting.
Teams choosing between Grok 4.3 and Grok 4.5 should weigh the tradeoff directly: Grok 4.3 keeps the larger 1 million token context window at Grok 4.20's original price point, while Grok 4.5 trades context size down to 500,000 tokens for a higher $2.00 input / $6.00 output price in exchange for coding-specific tuning and same-day availability inside Cursor and Grok Build.
Pricing
Standard API pricing is $1.25 per 1M input tokens / $2.50 per 1M output tokens for prompts under 200,000 tokens, per xAI's developer docs. Once a single request's prompt reaches 200,000 tokens, pricing doubles to $2.50 input / $5.00 output per 1M tokens, with cached input rising to $0.40 per 1M, a tier that matters given the model's 1 million token context window. The batch API takes a further 20% off the under-200K standard rates. The under-200K list price still matches xAI's published rate for Grok 4.20.
What a real job costs
| Job | Input | Output | Total |
|---|---|---|---|
| Summarise a 20-page PDF | $0.037 | $0.0025 | $0.040 |
| Support reply | $0.0025 | $0.0007 | $0.0032 |
| One coding agent run | $0.250 | $0.050 | $0.300 |
Budgets: 20-page PDF = 30k in / 1k out · Support reply = 2k in / 300 out · Coding agent run = 200k in / 20k out. Computed from the vendor's per-token prices at render time; cached-input discounts are not applied.
Key Features
- Configurable Reasoning With Native Tool Use: Supports adjustable reasoning effort alongside native function calling and structured JSON outputs, per xAI's developer documentation.
- 37 RPS / 10M TPM Rate Limits: Documented API limits of 37 requests per second and 10 million tokens per minute, among the highest published throughput ceilings in the Grok line.
- Live on AWS Bedrock and Azure AI Foundry: Reached AWS Bedrock through a Mantle inference engine and Microsoft Azure AI Foundry, which applies its own default Content Safety layer, within about two months of the April 2026 launch.
- Text-and-Image Input, Text-Only Output: Accepts text and image input and produces text-only output, per xAI's own docs; this contradicts secondary coverage claiming native video input, which xAI's documentation does not support.
- Cached-Input and Batch API Discounts: Reusing cached prompt context and routing through the batch API both cut cost versus standard per-token pricing, detailed in the pricing FAQ below.
Pros
- Posts this release's clearest verified benchmark gain against its immediate predecessor, unlike several other reported scores that lack independent confirmation.
- Its 10 million tokens-per-minute cap and 37-requests-per-second limit are among the most generous throughput ceilings xAI has published for any Grok model.
- Reached two major cloud platforms, AWS Bedrock and Azure AI Foundry, within about two months of launch, giving enterprise buyers deployment options beyond the direct xAI API.
Cons
- Unlike Grok 4, Grok 4 Fast, and Grok 4.1, this version ships without a published model card, leaving its safety and alignment methodology largely undocumented.
- Artificial Analysis's own launch-day and current Intelligence Index scores disagree, and neither xAI nor Artificial Analysis has publicly explained the gap.
- AA-Omniscience accuracy rose even as the model's non-hallucination rate fell on that same benchmark, a real quality tradeoff rather than a clean win.
Benchmarks
- GDPval-AA v2: 1,500 cited: Artificial Analysis · 07 Sep 2026 — Real knowledge-work deliverables judged against professionals, run by Artificial Analysis.
- AA Intelligence Index: 38 cited: Artificial Analysis · 07 Sep 2026 — Composite of 10 evaluations run by Artificial Analysis, 0 to 100.
- Output speed: 94 tok/s cited: Artificial Analysis · 07 Sep 2026 — Median tokens written per second as measured by Artificial Analysis.
A benchmark is an exam, not the job. Scores transfer unevenly between tasks, so weigh the one closest to your workload and read every figure with its source.
Frequently Asked Questions
What does Grok 4.3 actually cost?
Grok 4.3 costs $1.25 per 1 million input tokens and $2.50 per 1 million output tokens through xAI's direct API for prompts under 200,000 tokens, confirmed from xAI's developer documentation. Once a single request's prompt reaches 200,000 tokens, both prices double to $2.50 input and $5.00 output per 1 million tokens, with cached input rising to $0.40 per 1 million, a tier the model's 1 million token context window makes easy to hit on long documents or large codebases. The batch API applies a further 20% off the under-200K standard rates. The under-200K rate is identical to xAI's published rate for predecessor Grok 4.20, so Grok 4.3 did not raise or lower its base API list price despite the benchmark gains.
Can you use Grok 4.3 without paying?
No, Grok 4.3 has no free tier: access is entirely pay-as-you-go through xAI's API, billed per token, and reaching it requires an xAI developer account plus an API key. Managed access through AWS Bedrock or Microsoft Azure AI Foundry follows each platform's own billing terms rather than a separate free allowance from xAI. Budget-conscious teams can lower cost with the cached-input and batch-API discounts described above rather than expecting a free usage tier.
What are Grok 4.3's closest competitors?
Grok 4.3's closest competitors on hokai include Grok 4.20, its direct predecessor, priced the same but scoring lower on GDPval-AA; Grok 4.5, xAI's own coding-focused sibling that costs more per token for a smaller context window; and GPT-5.5, OpenAI's flagship, which still leads Grok 4.3 on the GDPval-AA evaluation. Teams choosing between xAI's own two current models should weigh context size against coding-specific tuning rather than price.
How does Grok 4.3 compare to Grok 4.20 in 2026?
Grok 4.3 posts a 321-point GDPval-AA Elo jump to 1500 over Grok 4.20's 1179, the clearest verified agentic-benchmark gain of this release, while API pricing stayed unchanged between the two versions. Grok 4.20 instead used a four-agent mixture-of-experts architecture with up to a 2 million token context on its multi-agent variant, a different design than the single Grok 4.3 model, whose own architecture xAI has not disclosed. Despite the GDPval-AA gain, Grok 4.3 still trails GPT-5.5 in that same evaluation, so the improvement over Grok 4.20 is real but does not close the gap with the market leader.
How do you set up Grok 4.3?
Getting started with Grok 4.3 means creating an xAI developer account, generating an API key, and calling the model by ID (grok-4.3, grok-4.3-latest, or the grok-latest alias) through an OpenAI-compatible endpoint. Teams already on AWS or Azure can instead reach it through Bedrock's Mantle inference engine or Microsoft Azure AI Foundry, using their existing cloud billing and IAM setup rather than a separate xAI account. Because reasoning effort is configurable, new integrations should start with a lower effort tier to gauge the roughly 22-second time-to-first-token on harder tasks before scaling up.
Top Alternatives
- Grok 4.20: Pick Grok 4.3 for the stronger verified agentic-benchmark score at an unchanged price; pick Grok 4.20 if its multi-agent variant's larger context ceiling matters more.
- Grok 4.5: Pick Grok 4.3 for the larger context window at a lower per-token price; pick Grok 4.5 for coding-specific tuning and same-day Cursor integration.
- GPT-5.5: Pick GPT-5.5 if you need the stronger GDPval-AA result today; pick Grok 4.3 if xAI ecosystem fit and context size matter more than closing that gap.