Reka: Flash 3

Reka: Flash 3

rekaai · Released Mar 12, 2025 Efficient
Intelligence #303 / 599
30.3 Our Score
Speed #142 / 281
95.0 tokens / sec
Input #197 / 599
$0.100 per 1M tokens
Output #185 / 599
$0.200 per 1M tokens
Context #396 / 599
65,536 tokens

Analysis Summary

Reka Flash 3 is a compact model from Reka AI with function calling and a 65K token context window. Its intelligence index of 4.1 is low relative to the broader field, and benchmark scores across reasoning, coding, and instruction following are below competitive mid-tier models.

The livecodebench score of 0.435 suggests some coding capability, and math performance is moderate. However, instruction-following scores are weak, which limits reliability for structured business outputs. There is no vision or tool use support beyond function calling.

At $0.10 input and $0.20 output, it is priced for high-volume, low-complexity tasks. Teams needing a cheap function-calling model for simple automation may find it adequate, but for anything requiring reliable reasoning or content quality, stronger alternatives exist at similar price points.

Assessed June 30, 2026

Editorial notes

Reka Flash 3 is a lightweight text model with function calling and a 65K context window, but its intelligence index of 4.1 and limited benchmark depth place it firmly in the mid-to-lower tier.

Rankings consider pricing, capabilities, benchmarks, and real-world applicability and are refreshed as new models launch. Feedback?

Performance Profile

Intelligence1.7Technical0Value7.5Content4
Intelligence 1.7/10
Technical 0/10
Content 4/10
Value 7.5/10

How Reka: Flash 3 compares

Reka: Flash 3 ranks #335 of 387 AI models we track for overall intelligence. Its 66K-token context window is larger than 34% of the models we list. At $0.10 per million input tokens it is cheaper than 67% of comparable models.

About Reka: Flash 3

Reka Flash 3 is a general-purpose, instruction-tuned large language model with 21 billion parameters, developed by Reka. It excels at general chat, coding tasks, instruction-following, and function calling. Featuring a..

Capabilities

Function Calling

Performance Indices

Source: Artificial Analysis

4.1 Intelligence Index
33.7 Math Index

Benchmark Scores

Intelligence

GPQA Diamond 52.9% Graduate-level scientific reasoning
HLE 5.1% Humanity's Last Exam
MMLU Pro 66.9% Multi-task language understanding
MATH 500 89.3% Mathematical problem-solving
AIME 51% Competition mathematics
AIME 2025 33.7% Competition mathematics (2025)
SciCode 26.7% Scientific computing

Technical

LiveCodeBench 43.5% Live coding evaluation

Content

IFBench 30.4% Instruction following

Benchmark data from Artificial Analysis and Hugging Face

How does Reka: Flash 3 stack up?

Compare side-by-side with other efficient models.

Compare Models

Model Information

OpenRouter ID rekaai/reka-flash-3
Providerrekaai
Release Date March 12, 2025
Context Length65,536 tokens
Max Completion65,536 tokens
Status Active

Pricing

Token Type Cost per 1M tokens Cost per 1K tokens
Input $0.10 $0.000100
Output $0.20 $0.000200

Live Performance

Live endpoint metrics, refreshed every 30 minutes.

165,776ms
Best Latency (TTFT)
17.5 tok/s
Best Throughput
0/1
Active Endpoints
Available via: Reka

Frequently asked questions about Reka: Flash 3

How much does Reka: Flash 3 cost?

Reka: Flash 3 costs $0.10 per million input tokens and $0.20 per million output tokens.

What is the context window of Reka: Flash 3?

Reka: Flash 3 has a context window of 65,536 tokens (66K).

What can Reka: Flash 3 do?

Reka: Flash 3 supports function calling.

Who created Reka: Flash 3?

Reka: Flash 3 is developed by Reka and was released on March 12, 2025.