Compare Try it out API Provider Benchmarks
\nMercury 2.5 is below average in intelligence, but well priced when comparing to other models of similar price. It's also notably fast and fairly concise. The model supports text input, outputs text, and has a 260k tokens context window.
\nMercury 2.5 scores 12 on the Artificial Analysis Intelligence Index, placing it below average among comparable models (median: 13). When evaluating the Intelligence Index, it generated 35M tokens, which is fairly concise in comparison to the median of 85M.
\nPricing for Mercury 2.5 is $0.25 per 1M input tokens (moderately priced, median: $0.25) and $0.75 per 1M output tokens (moderately priced, median: $0.90). On average, it costs $0.06 per task to evaluate Mercury 2.5 on the Intelligence Index.
\nAt 770 tokens per second, Mercury 2.5 is notably fast (109).
\nReasoning Yes This page shows the reasoning version of this model.
\nA non-reasoning variant may also exist.
\nInput modality Supports: text
\nOutput modality Supports: text
\nContext window 260k ~390 A4 pages of size 12 Arial font
\nMetrics are compared against models of the same class:
\nNon-reasoning models → compared only with other non-reasoning models
\nUpdated
\nArtificial Analysis Intelligence Index · Higher is better
\nOutput tokens per second · Higher is better
\nWeighted average cost (USD) per Intelligence Index task · Lower is better Intelligence Updated Capability Indexes Benchmarks Intelligence Index Comparisons Token Use Cost Context Window Speed Latency End-to-End Response Time Prompt Options
\nArtificial Analysis Intelligence Index v4.3.2 incorporates 10 evaluations: AA-Briefcase v1.1, GDPval-AA v2.1, AutomationBench-AA, Terminal-Bench 4.0, SciCode, Humanity's Last Exam, GDP.pdf, CritPt, AA-Omniscience, AA-LCR v1.1
\nArtificial Analysis Intelligence Index v4.3.2 includes: AA-Briefcase v1.1, GDPval-AA v2.1, AutomationBench-AA, Terminal-Bench 4.0, SciCode, Humanity's Last Exam, GDP.pdf, CritPt, AA-Omniscience, AA-LCR v1.1 . See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.
\nOpen Weights / Proprietary Reasoning / Non-Reasoning Text Only / Multimodal Inputs
\nArtificial Analysis Intelligence Index v4.3.2 incorporates 10 evaluations: AA-Briefcase v1.1, GDPval-AA v2.1, AutomationBench-AA, Terminal-Bench 4.0, SciCode, Humanity's Last Exam, GDP.pdf, CritPt, AA-Omniscience, AA-LCR v1.1 28 of 674 models Add model from specific provider Proprietary Open Weights Open Weights (Commercial Use Restricted)
\nArtificial Analysis Intelligence Index v4.3.2 includes: AA-Briefcase v1.1, GDPval-AA v2.1, AutomationBench-AA, Terminal-Bench 4.0, SciCode, Humanity's Last Exam, GDP.pdf, CritPt, AA-Omniscience, AA-LCR v1.1 . See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.
\nIndicates whether the model weights are available. Models are labelled as 'Commercial Use Restricted' if commercial use is limited by conditions, and as 'Non-commercial' if the license prohibits commercial use.
\nMeasures the performance of models on specific capabilities and industries
\nFinance & Accounting Strategy & Ops Legal Engineering Economics
\nIncorporates 7 evaluations: AA-Omniscience, GDPval-AA v2.1, AA-Briefcase v1.1, Humanity's Last Exam, AutomationBench-AA, AA-LCR v1.1, GDP.pdf · Higher is better 28 of 171 models Add model from specific provider
\nIntelligence evaluations measured independently by Artificial Analysis · Higher is better Coding Agentic Tool Use Private Dataset User Interaction Finance Medical Legal Intelligence Index Long Context Multimodal Instruction Following Faithfulness Writing Business See more 18 of 26 evaluations 28 of 674 models Add model from specific provider AA-Briefcase v1.1 Updated Agentic knowledge work, (Elo-500)/2000
\nGDPval-AA v2.1 Updated Agentic real-world work tasks, (Elo-500)/2000
\nAutomationBench-AA Updated Agentic SaaS workflows
\nTerminal-Bench 4.0 New Agentic coding & terminal use
\nSciCode Coding
\nHumanity's Last Exam Reasoning & knowledge
\nGDP.pdf New Professional document reasoning, All-pass
\nCritPt Under review Physics reasoning
\nAA-Omniscience Accuracy Knowledge
\nAA-Omniscience Non-Hallucination Rate 1 - hallucination rate
\nAA-LCR v1.1 Long context reasoning
\nHarvey LAB-AA Legal agentic work, criterion pass rate
\nEnterpriseOps-Gym-AA Agentic business operations
\nAA-AnalystAgent Quantitative analysis on spreadsheets & documents
\n𝜏³-Banking Agentic tool use
\nITBench-AA Kubernetes incident root-cause analysis
\nMMMU-Pro Visual reasoning
\nMLCR-AA New Medical long context reasoning
\nWhile model intelligence generally translates across use cases, specific evaluations may be more relevant for certain use cases.
\nArtificial Analysis Intelligence Index v4.3.2 includes: AA-Briefcase v1.1, GDPval-AA v2.1, AutomationBench-AA, Terminal-Bench 4.0, SciCode, Humanity's Last Exam, GDP.pdf, CritPt, AA-Omniscience, AA-LCR v1.1 . See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.
\nAA-Briefcase Elo AA-Briefcase Rubric Score (%) Analytical Quality & Presentation Elo
\nAA-Briefcase v1.1 is an agentic knowledge work benchmark developed by Artificial Analysis. AA-Briefcase Elo is a combined metric that aggregates rubric pass rate, analytical quality Elo and presentation Elo · Higher is better 28 of 192 models Add model from specific provider
\nAA-Briefcase Elo is a combined metric that aggregates analytical quality Elo, presentation Elo, and rubric pass rate, with rubric performance converted into Elo via synthetic head-to-head matches. Elo and 95% confidence interval bounds are clamped at 0.
\nAA-Omniscience Index AA-Omniscience Accuracy AA-Omniscience Hallucination Rate
\nAA-Omniscience Index (higher is better) measures knowledge reliability and hallucination. It rewards correct answers, penalizes hallucinations, and has no penalty for refusing to answer. Scores range from -100 to 100, where 0 means as many correct as incorrect answers, and negative scores mean more incorrect than correct. 28 of 549 models Add model from specific provider
\nAA-Omniscience Index (higher is better) measures knowledge reliability and hallucination. It rewards correct answers, penalizes hallucinations, and has no penalty for refusing to answer. Scores range from -100 to 100, where 0 means as many correct as incorrect answers, and negative scores mean more incorrect than correct.
\nIntelligence Index vs. Cost per Task Intelligence Index vs. Time per Task Intelligence Index vs. Output Speed Intelligence Index vs. End-to-End Response Time
\nArtificial Analysis Intelligence Index · Weighted average cost (USD) per Artificial Analysis Intelligence Index task 28 of 674 models Most attractive quadrant Pareto line Inception Xiaomi OpenAI Anthropic SpaceXAI Google Meta Z AI DeepSeek Alibaba MiniMax Multiverse Computing Tencent
\nWeighted average cost per Intelligence Index task. Each evaluation’s cost is calculated from input, cache hit, cache write, reasoning, and answer token prices, divided by task count, and weighted by its Intelligence Index weight.
\nArtificial Analysis Intelligence Index v4.3.2 includes: AA-Briefcase v1.1, GDPval-AA v2.1, AutomationBench-AA, Terminal-Bench 4.0, SciCode, Humanity's Last Exam, GDP.pdf, CritPt, AA-Omniscience, AA-LCR v1.1 . See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.
\nOutput Tokens per Task Intelligence Index vs. Output Tokens per Task Intelligence Index Token Use Intelligence Index vs. Token Use
\nWeighted average number of output tokens used to run one task in the Artificial Analysis Intelligence Index 28 of 674 models Answer Reasoning
\nThe number of tokens required per Intelligence Index task. This is calculated by multiplying the output tokens per eval by the relative weights of each benchmark in the Intelligence Index, then dividing by task count (excluding repeats).
\nCost per Task Intelligence Index vs. Cost per Task Evaluation Breakdown
\nWeighted average cost (USD) per Artificial Analysis Intelligence Index task, segmented by token type. Lower is better 28 of 674 models Answer Reasoning Cache Write Cache Hit Input
\nWeighted average cost per Intelligence Index task. Each evaluation’s cost is calculated from input, cache hit, cache write, reasoning, and answer token prices, divided by task count, and weighted by its Intelligence Index weight.
\nIntelligence Index Total Cost Intelligence Index vs. Total Cost
\nCost (USD) to run all evaluations in the Artificial Analysis Intelligence Index 28 of 674 models Add model from specific provider Output Reasoning Cache Write Cache Read Non-Cache Input
\nThe cost to run the evaluations in the Artificial Analysis Intelligence Index, calculated using the model's input, cache hit, cache write, reasoning, and answer token prices and the number of tokens used across evaluations (excluding repeats).
\nCache Hit, Input, and Output Pricing Blended Price Blended Price (Stacked) Cache Discount Intelligence Index vs. Price Intelligence Index vs. Price (Log, Inverted) Image Input Pricing
\nPrice (USD per M Tokens) 28 of 674 models Add model from specific provider Cache Hit Input Output
\nPrice per token for cached prompts (previously processed), typically offering a significant discount compared to regular input price, represented as USD per million tokens. The values shown here are the cache hit price; cache write and cache storage are billed separately and vary by provider — see "Cache pricing by provider" for detail.
\n4 more notes
\nContext Window Intelligence Index vs. Context Window
\nContext window: tokens limit · Higher is better 28 of 674 models Add model from specific provider
\nLarger context windows are relevant to RAG (Retrieval Augmented Generation) LLM workflows which typically involve reasoning and information retrieval of large amounts of data.
\nMaximum number of combined input & output tokens. Output tokens commonly have a significantly lower limit (varied by model).
\nMeasured by Output Speed (tokens per second)
\nOutput Speed Output Speed by Prompt Type Output Speed Variance Output Speed Over Time Output Speed vs. Price Latency vs. Output Speed
\nOutput tokens per second · Higher is better 28 of 674 models Add model from specific provider
\nTokens per second received while the model is generating tokens (ie. after first chunk has been received from the API for models which support streaming).
\nFigures represent performance of the model's first-party API or the median across providers where a first-party API is not available.
\nTime per Task Intelligence Index vs. Time per Task Cost vs. Time per Task
\nWeighted average decode time (minutes) per task; excludes TTFT and overhead time · Lower is better 28 of 674 models
\nThe weighted average time (seconds) per Artificial Analysis Intelligence Index task. This is calculated by dividing output tokens per task by output speed, weighted by the relative weights of each benchmark in the Intelligence Index.
\nMeasured by Time (seconds) to First Token
\nTime To First Answer Token Time To First Token Latency by Prompt Type Latency Variance Latency Over Time
\nSeconds to first answer token received · Accounts for reasoning model 'thinking' time 28 of 674 models Add model from specific provider Thinking (reasoning models, when applicable) Input processing
\nTime to first answer token received, in seconds, after API request sent. For reasoning models, this includes the 'thinking' time of the model before providing an answer. For models which do not support streaming, this represents time to receive the completion.
\nSeconds to output 500 tokens, calculated based on time to first token, 'thinking' time for reasoning models, and output speed
\nEnd-to-End Response Time End-to-End Response Time by Prompt Type End-to-End Response Time Over Time
\nSeconds to output 500 tokens, including reasoning model 'thinking' time · Lower is better 28 of 674 models Add model from specific provider Outputting time 'Thinking' time (reasoning models) Input processing time
\nSeconds to receive a 500 token response. Key components:
\nInput time: Time to receive the first response token
\nFigures represent performance of the model's first-party API or the median across providers where a first-party API is not available.
\nCommon questions about Mercury 2.5
\nMercury 2.5 was released on September 8, 2026.
\nMercury 2.5 was created by Inception.
\nMercury 2.5 scores 12 on the Artificial Analysis Intelligence Index, placing it below average among other reasoning models in a similar price tier (median: 13).
\nMercury 2.5 generates output at 770.4 tokens per second (based on Inception's API), which is well above average compared to other reasoning models in a similar price tier (median: 108.6 t/s).
\nMercury 2.5 has a time to first token (TTFT) of 2.95s (based on Inception's API), which is somewhat higher than average compared to other reasoning models in a similar price tier (median: 2.23s).
\nMercury 2.5 costs $0.25 per 1M input tokens (better than average, median: $0.25) and $0.75 per 1M output tokens (better than average, median: $0.90), based on Inception's API.
\nMercury 2.5 costs $0.25 per 1M input tokens and $0.75 per 1M output tokens (based on Inception's API). For a blended rate (7:2:1 cache hit/input/output ratio), this is $0.14 per 1M tokens. Pricing may vary by provider. Compare provider pricing
\nWhen evaluated on the Intelligence Index, Mercury 2.5 generated 35M output tokens, which is better than average compared to other reasoning models in a similar price tier (median: 85M).
\nYes, Mercury 2.5 is a reasoning model. It uses extended thinking or chain-of-thought reasoning to work through complex problems before providing an answer.
\nMercury 2.5 supports text input.
\nMercury 2.5 supports text output.
\nNo, Mercury 2.5 does not support image input. It can only process text.
\nNo, Mercury 2.5 is not multimodal. It only supports text input.
\nMercury 2.5 has a context window of 260k tokens. This determines how much text and conversation history the model can process in a single request.
\nNo, Mercury 2.5 is proprietary. The model weights are not publicly available.
\nMercury 2.5 is a proprietary model and Inception has not disclosed the model size or parameter count.
\nMercury 2.5 achieves a score of 12 on the Artificial Analysis Intelligence Index. This composite benchmark evaluates models across reasoning, knowledge, mathematics, and coding.
\nYes, Mercury 2.5 is available via API through 1 provider. Compare API providers
\nMercury 2.5 is available through 1 API provider. Compare providers
\n\nGet notified about new articles
\nEmail address Subscribe Artificial Analysis
\nExplore
\n