Tech Meridian← ALL MODELS
PROMY MERIDIAN

XAI · MODEL RELEASE TRACKER

Grok 4.7

Grok 4.7 is a version of xAI's Grok model reported as politically right-leaning and used as the judge model in a Nous Research evaluation of propaganda in Chinese models.

CURRENT SNAPSHOT4/5 DIMENSIONS WITH DATA

The dimensions that change the decision.

PRICING

Not established from the available sources.

CONTEXT WINDOW
Experiment token allowanceDuring the experiment, each agent (including Grok 4.7) was allowed up to 40 million tokens and up to twelve simulated weeks.
MODALITIES
ModalityEvaluated as a large language model (LLM) agent in multi-agent economic experiments.
BENCHMARKS
CursorBench 4.0 price-performancexAI reports Grok 4.7 is at the frontier in price-performance on CursorBench 4.0, a benchmark that stresses longer-running coding tasks.
GDPval and AA Briefcase performancexAI reports Grok 4.7 improves on Grok 4.6 on GDPval and AA Briefcase (benchmarks for professional tasks) and performs comparably to other frontier models.
Artificial Analysis Intelligence Index (v4.3.2) scoreAn independent report gives Grok 4.7 a combined score of 46 on the Artificial Analysis Intelligence Index (v4.3.2), placing it mid-pack versus competitors (Claude Fable 5.1 and GPT-6 scored 53).
Terminal-Bench 4.0 (agentic coding)Reported Terminal-Bench 4.0 agentic-coding score of 26% (lower than certain competitor models reported in the same article).
Reported benchmark gainsxAI reports gains across published evaluations, including software engineering benchmarks (CursorBench, DeepSWE), multi-hour terminal and office work (Terminal-Bench, AA Briefcase), and domain benchmarks such as EEBench for electrical engineering and evaluations for legal work.
Agent evaluation on three long-horizon tasksIn experiments reported on Amazon Bedrock, an agent using Grok 4.7 answered 67 of 72 runs correctly across three search-oriented tasks; at the 'xhigh' reasoning effort it answered 18/18 correctly, and at 'low' it answered 15/18 while using about one-eighth of the output tokens.
Included in Vals AI refund-agent experimentGrok 4.7 was one of six AI models evaluated by Vals AI and researcher Matilda Glynn-Henley in an experiment that tasked agents with securing a £350 train refund.
Political candidate recommendation bias in studyIn a study asking models to name one candidate for upcoming races, Grok 4.7 recommended the Republican candidate in 92% of its answers.
Agreement with GPT-6.1 SolAcross nearly 500 races in the study, Grok 4.7 and GPT-6.1 Sol agreed on the recommended candidate less than 10% of the time.
Artificial Analysis Intelligence Index points changeGrok 4.7 lost 12 points in Artificial Analysis's overhauled Intelligence Index (version 5).
ReplayRCA benchmark — causal repair on 62 calibrated systematic failuresReported as the best of 13 frontier models; causal repair = +51.3 points on 62 calibrated systematic failures. (The paper notes the top eight models are statistically indistinguishable.)
Double-auction welfare under full disclosureIn the study's double-auction experiments, full information disclosure inflated price demands and lowered Grok 4.7's welfare from 97% to 62%.
Used as judge model in Nous Research evaluationServed as the judge model in Nous Research's evaluation of propaganda in Chinese models, checking whether responses mention, omit, or contradict facts.
AVAILABILITY
Accessed via API in studyResearchers queried Grok 4.7 using its API to obtain answers for the study.

RELEASE TIMELINE

Published, source-backed release events only.

1 SOURCES · IMPORTANCE 7.0

Artemis Security launches Orion-1, an AI to investigate and respond to corporate cyberattacks

Artemis Security has launched Orion-1, a purpose-built AI model that ingests logs from endpoints, networks, cloud services and SaaS, reconstructs attacks, assigns confidence scores, and can recommend or initiate responses; the company says Orion-1 outperformed general-purpose models like Claude Opus 5.5 and GPT-6 Sol on its internal Decision-Grade Readiness (DGR) benchmark. Orion-1 is in private preview; Artemis says no customer data was used for training and it is unclear whether Orion-1 was built from scratch or fine-tuned from another foundation model.

→
2 SOURCES · IMPORTANCE 8.0

SpaceXAI releases Grok 4.7 — faster, cheaper model for coding and knowledge work

SpaceXAI released Grok 4.7, a new base-model upgrade positioned as its most capable model for coding and knowledge work; it reportedly uses a larger base model, longer RL training focused on long-running tasks, and an entirely new safeguard stack. The company says Grok 4.7 matches the price and speed of Grok 4.6 while improving verification, longer-context handling, document/presentation generation, benchmark performance (CursorBench 4.0, GDPval, AA Briefcase), and safety metrics (LatchBio biosafety 62.4%, HackerBench v0.3 allowing 3.3% risky prompts); it is available in Cursor, Grok Build, the Grok API, third-party harnesses and cloud platforms, priced from $2 per million input tokens and $6 per million output tokens, with an optional faster (2x) variant at double the price.

→

VERIFIABLE FACTS

Every value stays attached to a source and date.

BENCHMARKS · CursorBench 4.0 price-performanceDEVELOPER CLAIM

xAI reports Grok 4.7 is at the frontier in price-performance on CursorBench 4.0, a benchmark that stresses longer-running coding tasks.

BENCHMARKS · GDPval and AA Briefcase performanceDEVELOPER CLAIM

xAI reports Grok 4.7 improves on Grok 4.6 on GDPval and AA Briefcase (benchmarks for professional tasks) and performs comparably to other frontier models.

CAPABILITIES · Grok Bot harness and conversational gainsDEVELOPER CLAIM

xAI reports Grok 4.7 was trained to natively understand the Grok Bot harness, which the company credits for improvements in conversational tasks and general knowledge work.

BENCHMARKS · Reported benchmark gainsDEVELOPER CLAIM

xAI reports gains across published evaluations, including software engineering benchmarks (CursorBench, DeepSWE), multi-hour terminal and office work (Terminal-Bench, AA Briefcase), and domain benchmarks such as EEBench for electrical engineering and evaluations for legal work.

BENCHMARKS · Agent evaluation on three long-horizon tasksDEVELOPER CLAIM

In experiments reported on Amazon Bedrock, an agent using Grok 4.7 answered 67 of 72 runs correctly across three search-oriented tasks; at the 'xhigh' reasoning effort it answered 18/18 correctly, and at 'low' it answered 15/18 while using about one-eighth of the output tokens.

WHAT CHANGED

Stored passport versions, without reconstructed history.

Passport updated29 facts
SAFETY · Consumer app refused candidate recommendation− The consumer chatbot apps refused to answer the forced 'Who should I vote for' prompt, while the model when accessed via API returned recommendations.
SAFETY · Political leaning / bias+ Reported as politically right-leaning.
BENCHMARKS · Used as judge model in Nous Research evaluation+ Served as the judge model in Nous Research's evaluation of propaganda in Chinese models, checking whether responses mention, omit, or contradict facts.
Passport updated28 facts
LIMITATIONS · Less balanced bargaining outcomes with exact information+ In bargaining experiments, providing exact information led Grok 4.7 to produce less balanced agreements.
BENCHMARKS · Double-auction welfare under full disclosure+ In the study's double-auction experiments, full information disclosure inflated price demands and lowered Grok 4.7's welfare from 97% to 62%.
CAPABILITIES · Evaluated tasks+ Included in experiments on multi-resource trade, bilateral bargaining, and double auctions.
MODALITIES · Modality+ Evaluated as a large language model (LLM) agent in multi-agent economic experiments.
Passport updated24 facts
BENCHMARKS · ReplayRCA benchmark — causal repair on 62 calibrated systematic failures+ Reported as the best of 13 frontier models; causal repair = +51.3 points on 62 calibrated systematic failures. (The paper notes the top eight models are statistically indistinguishable.)
Passport updated23 facts
BENCHMARKS · Artificial Analysis Intelligence Index points change+ Grok 4.7 lost 12 points in Artificial Analysis's overhauled Intelligence Index (version 5).
Passport updated22 facts
BENCHMARKS · Agreement with GPT-6.1 Sol+ Across nearly 500 races in the study, Grok 4.7 and GPT-6.1 Sol agreed on the recommended candidate less than 10% of the time.
AVAILABILITY · Accessed via API in study+ Researchers queried Grok 4.7 using its API to obtain answers for the study.
AVAILABILITY · Availability on Amazon Bedrock− Grok 4.7 is available on Amazon Bedrock (reported by AWS).
SAFETY · Consumer app refused candidate recommendation+ The consumer chatbot apps refused to answer the forced 'Who should I vote for' prompt, while the model when accessed via API returned recommendations.
BENCHMARKS · Political candidate recommendation bias in study+ In a study asking models to name one candidate for upcoming races, Grok 4.7 recommended the Republican candidate in 92% of its answers.
SAFETY · Self-verification behavior− The model checks and verifies its own output more carefully before moving on; xAI states this self-verification helps the model fail less catastrophically on long trajectories.
Passport updated20 facts
CONTEXT WINDOW · Experiment token allowance+ During the experiment, each agent (including Grok 4.7) was allowed up to 40 million tokens and up to twelve simulated weeks.
CONTEXT WINDOW · Improved management of longer context− SpaceXAI reports the model is better at managing longer context compared with Grok 4.6.
BENCHMARKS · Included in Vals AI refund-agent experiment+ Grok 4.7 was one of six AI models evaluated by Vals AI and researcher Matilda Glynn-Henley in an experiment that tasked agents with securing a £350 train refund.
CAPABILITIES · Web browsing and email-sending access (in experiment)+ In the experiment, Grok 4.7 was tested with web browsing capabilities and the ability to send emails autonomously.
Passport updated18 facts
BENCHMARKS · Agent evaluation on three long-horizon tasks+ In experiments reported on Amazon Bedrock, an agent using Grok 4.7 answered 67 of 72 runs correctly across three search-oriented tasks; at the 'xhigh' reasoning effort it answered 18/18 correctly, and at 'low' it answered 15/18 while using about one-eighth of the output tokens.
AVAILABILITY · Availability on Amazon Bedrock+ Grok 4.7 is available on Amazon Bedrock (reported by AWS).
AVAILABILITY · Availability and APIs− Grok 4.7 is available on Amazon Bedrock, served on the bedrock-runtime endpoint through cross-Region inference profiles, and supports the Responses, Chat Completions, and Converse APIs.
CONTEXT WINDOW · Context window− Grok 4.7 offers a 500K token context window.
CONTEXT WINDOW · Improved management of longer context+ SpaceXAI reports the model is better at managing longer context compared with Grok 4.6.
RELEASE · Base model and training changes vs Grok 4.6+ According to SpaceXAI, Grok 4.7 uses a larger base model than Grok 4.6 and was trained with a longer reinforcement-learning run on a harder mix of tasks weighted toward problems that take many hours to complete.
CAPABILITIES · Reasoning-effort setting+ Grok 4.7 exposes a reasoning effort setting that controls how much deliberation happens before each response.
CAPABILITIES · Self-verification and longer-task capability+ According to SpaceXAI, Grok 4.7 is trained to work longer on more difficult tasks and "knows when to check its own work."
RELEASE · Training and base model− xAI reports Grok 4.7 uses a new, larger base model and was trained with a longer reinforcement-learning run over a harder mix of tasks, deliberately weighted toward problems that take many hours to complete.
Passport updated15 facts
AVAILABILITY · Availability and APIs+ Grok 4.7 is available on Amazon Bedrock, served on the bedrock-runtime endpoint through cross-Region inference profiles, and supports the Responses, Chat Completions, and Converse APIs.
CAPABILITIES · Configurable reasoning effort+ The model supports configurable reasoning effort at four levels: low, medium, high, and xhigh.
CONTEXT WINDOW · Context window+ Grok 4.7 offers a 500K token context window.
CAPABILITIES · Grok Bot harness and conversational gains+ xAI reports Grok 4.7 was trained to natively understand the Grok Bot harness, which the company credits for improvements in conversational tasks and general knowledge work.
CONTEXT WINDOW · context handling− Described by xAI as better at managing longer context than prior versions (no numeric context-window size provided).
CAPABILITIES · Primary use cases+ xAI positions Grok 4.7 as built for coding, long-running agents, and professional knowledge work (e.g., legal, medical, financial workflows).
BENCHMARKS · Reported benchmark gains+ xAI reports gains across published evaluations, including software engineering benchmarks (CursorBench, DeepSWE), multi-hour terminal and office work (Terminal-Bench, AA Briefcase), and domain benchmarks such as EEBench for electrical engineering and evaluations for legal work.
SAFETY · Self-verification behavior+ The model checks and verifies its own output more carefully before moving on; xAI states this self-verification helps the model fail less catastrophically on long trajectories.
RELEASE · Training and base model+ xAI reports Grok 4.7 uses a new, larger base model and was trained with a longer reinforcement-learning run over a harder mix of tasks, deliberately weighted toward problems that take many hours to complete.