AI models for Contract analysis / clause extraction
Contract analysis locates the clauses that matter in an agreement — termination, liability cap, indemnity, renewal, governing law, change of control — and returns each one with its text, its position and the obligation or date it creates.
The value is in the aggregate. Once a portfolio of agreements has been read this way, questions that previously required a lawyer to open every file become queries: which contracts auto-renew this quarter, which cap liability below our threshold, which require notice to a counterparty on a change of ownership. Accuracy requirements are high and the material is long, so the working pattern is retrieval followed by careful reading: find the candidate clauses first, then have a capable model read only those.
Models in this group take text or whole documents as input: plain text, PDFs, scanned pages and office files. Each specification table gives three hardware tiers — Minimum, the smallest setup on which the model runs correctly; Medium, the usual production configuration; and High, a configuration sized for peak volume. The rate is the number of sample inputs processed per hour on that hardware. Use these rates for initial sizing. Before production, benchmark your own data to validate accuracy, latency, throughput and cost. The sample input here is one A4 page of contract text, about 400 words.
Llama 3.3 70B
Vendor: Meta
What it does: reads clause text and states what it means in practice — the notice period, the cap, the trigger — with the clause quoted as its source. The most reliable option for material that will inform a legal position.
Requirement
Minimum
Medium
High
GPU type
2× RTX 4090 (reduced precision)
H100 80 GB
4× H100 80 GB
VRAM
48 GB combined
80 GB
320 GB combined
vCPUs
16
24
64
RAM
64 GB
128 GB
512 GB
Server
2× RTX 4090 24 GB
1× H100 SXM 80 GB
4× H100 SXM 80 GB
Rate (pages/hour)
≈ 250
≈ 1,100
≈ 4,500
Qwen2.5 32B
Vendor: Alibaba Cloud
What it does: reads a whole agreement in one pass, which matters because clauses cross-reference each other and definitions sit tens of pages away from where they are used.
Requirement
Minimum
Medium
High
GPU type
RTX 4090 (reduced precision)
L40S 48 GB
2× H100 80 GB
VRAM
22 GB
48 GB
160 GB combined
vCPUs
12
16
48
RAM
48 GB
64 GB
256 GB
Server
1× RTX 4090 24 GB
1× L40S 48 GB
2× H100 SXM 80 GB
Rate (pages/hour)
≈ 350
≈ 1,100
≈ 4,200
Mistral Small 3
Vendor: Mistral AI
What it does: a compact model for first-pass reading across a large portfolio, flagging the agreements that need a closer look rather than analysing each in depth.
Requirement
Minimum
Medium
High
GPU type
RTX 4090
L40S 48 GB
H100 80 GB
VRAM
24 GB
48 GB
80 GB
vCPUs
12
16
32
RAM
48 GB
64 GB
128 GB
Server
1× RTX 4090 24 GB
1× L40S 48 GB
1× H100 SXM 80 GB
Rate (pages/hour)
≈ 700
≈ 2,000
≈ 6,000
BGE-M3
Vendor: Beijing Academy of Artificial Intelligence
What it does: fingerprints every clause in the portfolio so that clauses of the same kind can be found across agreements regardless of how they are worded or titled.
Requirement
Minimum
Medium
High
GPU type
RTX 3060
RTX 4090
A100 80 GB
VRAM
8 GB
24 GB
80 GB
vCPUs
6
12
24
RAM
16 GB
32 GB
64 GB
Server
1× RTX 3060 12 GB
1× RTX 4090 24 GB
1× A100 SXM 80 GB
Rate (pages/hour)
≈ 6,000
≈ 30,000
≈ 90,000
BGE Reranker v2-M3
Vendor: Beijing Academy of Artificial Intelligence
What it does: re-sorts candidate clauses by how well they actually match the clause type sought, which is what keeps a reading model from being handed the wrong paragraph.
Requirement
Minimum
Medium
High
GPU type
RTX 3090
RTX 4090
A100 80 GB
VRAM
12 GB
24 GB
80 GB
vCPUs
8
12
24
RAM
32 GB
48 GB
96 GB
Server
1× RTX 3090 24 GB
1× RTX 4090 24 GB
1× A100 SXM 80 GB
Rate (clause-query pairs/hour)
≈ 40,000
≈ 150,000
≈ 400,000
DeBERTa v3
Vendor: Microsoft
What it does: fitted to your clause taxonomy from examples marked up by your own lawyers, so it classifies clauses the way your practice does. The cheapest option per page once that markup exists.
Requirement
Minimum
Medium
High
GPU type
RTX 3060
RTX 4090
A100 80 GB
VRAM
6 GB
24 GB
80 GB
vCPUs
4
8
24
RAM
16 GB
32 GB
64 GB
Server
1× RTX 3060 12 GB
1× RTX 4090 24 GB
1× A100 SXM 80 GB
Rate (pages/hour)
≈ 8,000
≈ 30,000
≈ 90,000
GLiNER
Vendor: Urchade Zaratiana and contributors
What it does: extracts the specific values inside a clause — dates, notice periods, monetary caps, jurisdictions — from a written list, with no training data required.
Requirement
Minimum
Medium
High
GPU type
RTX 3060
RTX 4090
A100 80 GB
VRAM
6 GB
24 GB
80 GB
vCPUs
4
8
24
RAM
16 GB
32 GB
64 GB
Server
1× RTX 3060 12 GB
1× RTX 4090 24 GB
1× A100 SXM 80 GB
Rate (pages/hour)
≈ 3,000
≈ 12,000
≈ 34,000
Choosing between them
Contract work sets a high bar for precision and demands that every answer cite its source clause. Our consultants review a sample of your agreements and your clause list, then recommend the retrieval and reading combination, with the review process that keeps a lawyer in control of what the system asserts.
At the start of a project we may run a short proof of concept on a sample of your own data, measuring the accuracy and the throughput the model actually achieves on your material. That replaces the estimates on this page with real figures, so the cost and the schedule for the full engagement are known before it is committed.
Share a representative sample, expected volume, latency target and deployment location for Contract analysis / clause extraction. We benchmark the shortlisted models, recommend the lowest-cost GPU configuration that meets the target, and scale it from pilot capacity to a dedicated production cluster.